src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_avx_256_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS avx_256_single kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_avx_256_single.h"
  50 #include "kernelutil_x86_avx_256_single.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_VF_avx_256_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            LJEwald
  56  * Geometry:                   Particle-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_VF_avx_256_single
  61                     (t_nblist                    * gmx_restrict       nlist,
  62                      rvec                        * gmx_restrict          xx,
  63                      rvec                        * gmx_restrict          ff,
  64                      t_forcerec                  * gmx_restrict          fr,
  65                      t_mdatoms                   * gmx_restrict     mdatoms,
  66                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  67                      t_nrnb                      * gmx_restrict        nrnb)
  68 {
  69     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  70      * just 0 for non-waters.
  71      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
  72      * jnr indices corresponding to data put in the four positions in the SIMD register.
  73      */
  74     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  75     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  76     int              jnrA,jnrB,jnrC,jnrD;
  77     int              jnrE,jnrF,jnrG,jnrH;
  78     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  79     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  80     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  81     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
  82     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  83     real             rcutoff_scalar;
  84     real             *shiftvec,*fshift,*x,*f;
  85     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
  86     real             scratch[4*DIM];
  87     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  88     real *           vdwioffsetptr0;
  89     real *           vdwgridioffsetptr0;
  90     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  91     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
  92     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  93     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  94     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
  95     real             *charge;
  96     int              nvdwtype;
  97     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  98     int              *vdwtype;
  99     real             *vdwparam;
 100     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 101     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 102     __m256           c6grid_00;
 103     real             *vdwgridparam;
 104     __m256           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 105     __m256           one_half  = _mm256_set1_ps(0.5);
 106     __m256           minus_one = _mm256_set1_ps(-1.0);
 107     __m256i          ewitab;
 108     __m128i          ewitab_lo,ewitab_hi;
 109     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 110     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 111     real             *ewtab;
 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     vdwgridparam     = fr->ljpme_c6grid;
 133     sh_lj_ewald      = _mm256_set1_ps(fr->ic->sh_lj_ewald);
 134     ewclj            = _mm256_set1_ps(fr->ewaldcoeff_lj);
 135     ewclj2           = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
 136
 137     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 138     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 139     beta2            = _mm256_mul_ps(beta,beta);
 140     beta3            = _mm256_mul_ps(beta,beta2);
 141
 142     ewtab            = fr->ic->tabq_coul_FDV0;
 143     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 144     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 145
 146     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 147     rcutoff_scalar   = fr->rcoulomb;
 148     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 149     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 150
 151     sh_vdw_invrcut6  = _mm256_set1_ps(fr->ic->sh_invrc6);
 152     rvdw             = _mm256_set1_ps(fr->rvdw);
 153
 154     /* Avoid stupid compiler warnings */
 155     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 156     j_coord_offsetA = 0;
 157     j_coord_offsetB = 0;
 158     j_coord_offsetC = 0;
 159     j_coord_offsetD = 0;
 160     j_coord_offsetE = 0;
 161     j_coord_offsetF = 0;
 162     j_coord_offsetG = 0;
 163     j_coord_offsetH = 0;
 164
 165     outeriter        = 0;
 166     inneriter        = 0;
 167
 168     for(iidx=0;iidx<4*DIM;iidx++)
 169     {
 170         scratch[iidx] = 0.0;
 171     }
 172
 173     /* Start outer loop over neighborlists */
 174     for(iidx=0; iidx<nri; iidx++)
 175     {
 176         /* Load shift vector for this list */
 177         i_shift_offset   = DIM*shiftidx[iidx];
 178
 179         /* Load limits for loop over neighbors */
 180         j_index_start    = jindex[iidx];
 181         j_index_end      = jindex[iidx+1];
 182
 183         /* Get outer coordinate index */
 184         inr              = iinr[iidx];
 185         i_coord_offset   = DIM*inr;
 186
 187         /* Load i particle coords and add shift vector */
 188         gmx_mm256_load_shift_and_1rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 189
 190         fix0             = _mm256_setzero_ps();
 191         fiy0             = _mm256_setzero_ps();
 192         fiz0             = _mm256_setzero_ps();
 193
 194         /* Load parameters for i particles */
 195         iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
 196         vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 197         vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
 198
 199         /* Reset potential sums */
 200         velecsum         = _mm256_setzero_ps();
 201         vvdwsum          = _mm256_setzero_ps();
 202
 203         /* Start inner kernel loop */
 204         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 205         {
 206
 207             /* Get j neighbor index, and coordinate index */
 208             jnrA             = jjnr[jidx];
 209             jnrB             = jjnr[jidx+1];
 210             jnrC             = jjnr[jidx+2];
 211             jnrD             = jjnr[jidx+3];
 212             jnrE             = jjnr[jidx+4];
 213             jnrF             = jjnr[jidx+5];
 214             jnrG             = jjnr[jidx+6];
 215             jnrH             = jjnr[jidx+7];
 216             j_coord_offsetA  = DIM*jnrA;
 217             j_coord_offsetB  = DIM*jnrB;
 218             j_coord_offsetC  = DIM*jnrC;
 219             j_coord_offsetD  = DIM*jnrD;
 220             j_coord_offsetE  = DIM*jnrE;
 221             j_coord_offsetF  = DIM*jnrF;
 222             j_coord_offsetG  = DIM*jnrG;
 223             j_coord_offsetH  = DIM*jnrH;
 224
 225             /* load j atom coordinates */
 226             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 227                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 228                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 229                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 230                                                  &jx0,&jy0,&jz0);
 231
 232             /* Calculate displacement vector */
 233             dx00             = _mm256_sub_ps(ix0,jx0);
 234             dy00             = _mm256_sub_ps(iy0,jy0);
 235             dz00             = _mm256_sub_ps(iz0,jz0);
 236
 237             /* Calculate squared distance and things based on it */
 238             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 239
 240             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 241
 242             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 243
 244             /* Load parameters for j particles */
 245             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 246                                                                  charge+jnrC+0,charge+jnrD+0,
 247                                                                  charge+jnrE+0,charge+jnrF+0,
 248                                                                  charge+jnrG+0,charge+jnrH+0);
 249             vdwjidx0A        = 2*vdwtype[jnrA+0];
 250             vdwjidx0B        = 2*vdwtype[jnrB+0];
 251             vdwjidx0C        = 2*vdwtype[jnrC+0];
 252             vdwjidx0D        = 2*vdwtype[jnrD+0];
 253             vdwjidx0E        = 2*vdwtype[jnrE+0];
 254             vdwjidx0F        = 2*vdwtype[jnrF+0];
 255             vdwjidx0G        = 2*vdwtype[jnrG+0];
 256             vdwjidx0H        = 2*vdwtype[jnrH+0];
 257
 258             /**************************
 259              * CALCULATE INTERACTIONS *
 260              **************************/
 261
 262             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 263             {
 264
 265             r00              = _mm256_mul_ps(rsq00,rinv00);
 266
 267             /* Compute parameters for interactions between i and j atoms */
 268             qq00             = _mm256_mul_ps(iq0,jq0);
 269             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 270                                             vdwioffsetptr0+vdwjidx0B,
 271                                             vdwioffsetptr0+vdwjidx0C,
 272                                             vdwioffsetptr0+vdwjidx0D,
 273                                             vdwioffsetptr0+vdwjidx0E,
 274                                             vdwioffsetptr0+vdwjidx0F,
 275                                             vdwioffsetptr0+vdwjidx0G,
 276                                             vdwioffsetptr0+vdwjidx0H,
 277                                             &c6_00,&c12_00);
 278
 279             c6grid_00       = gmx_mm256_load_8real_swizzle_ps(vdwgridioffsetptr0+vdwjidx0A,
 280                                                                   vdwgridioffsetptr0+vdwjidx0B,
 281                                                                   vdwgridioffsetptr0+vdwjidx0C,
 282                                                                   vdwgridioffsetptr0+vdwjidx0D,
 283                                                                   vdwgridioffsetptr0+vdwjidx0E,
 284                                                                   vdwgridioffsetptr0+vdwjidx0F,
 285                                                                   vdwgridioffsetptr0+vdwjidx0G,
 286                                                                   vdwgridioffsetptr0+vdwjidx0H);
 287
 288             /* EWALD ELECTROSTATICS */
 289
 290             /* Analytical PME correction */
 291             zeta2            = _mm256_mul_ps(beta2,rsq00);
 292             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 293             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 294             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 295             felec            = _mm256_mul_ps(qq00,felec);
 296             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 297             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 298             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv00,sh_ewald),pmecorrV);
 299             velec            = _mm256_mul_ps(qq00,velec);
 300
 301             /* Analytical LJ-PME */
 302             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 303             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 304             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 305             exponent         = gmx_simd_exp_r(ewcljrsq);
 306             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 307             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 308             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 309             vvdw6            = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
 310             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 311             vvdw             = _mm256_sub_ps(_mm256_mul_ps( _mm256_sub_ps(vvdw12 , _mm256_mul_ps(c12_00,_mm256_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
 312                                           _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_add_ps(_mm256_mul_ps(c6_00,sh_vdw_invrcut6),_mm256_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
 313             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 314             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
 315
 316             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 317
 318             /* Update potential sum for this i atom from the interaction with this j atom. */
 319             velec            = _mm256_and_ps(velec,cutoff_mask);
 320             velecsum         = _mm256_add_ps(velecsum,velec);
 321             vvdw             = _mm256_and_ps(vvdw,cutoff_mask);
 322             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 323
 324             fscal            = _mm256_add_ps(felec,fvdw);
 325
 326             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 327
 328             /* Calculate temporary vectorial force */
 329             tx               = _mm256_mul_ps(fscal,dx00);
 330             ty               = _mm256_mul_ps(fscal,dy00);
 331             tz               = _mm256_mul_ps(fscal,dz00);
 332
 333             /* Update vectorial force */
 334             fix0             = _mm256_add_ps(fix0,tx);
 335             fiy0             = _mm256_add_ps(fiy0,ty);
 336             fiz0             = _mm256_add_ps(fiz0,tz);
 337
 338             fjptrA             = f+j_coord_offsetA;
 339             fjptrB             = f+j_coord_offsetB;
 340             fjptrC             = f+j_coord_offsetC;
 341             fjptrD             = f+j_coord_offsetD;
 342             fjptrE             = f+j_coord_offsetE;
 343             fjptrF             = f+j_coord_offsetF;
 344             fjptrG             = f+j_coord_offsetG;
 345             fjptrH             = f+j_coord_offsetH;
 346             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,tx,ty,tz);
 347
 348             }
 349
 350             /* Inner loop uses 145 flops */
 351         }
 352
 353         if(jidx<j_index_end)
 354         {
 355
 356             /* Get j neighbor index, and coordinate index */
 357             jnrlistA         = jjnr[jidx];
 358             jnrlistB         = jjnr[jidx+1];
 359             jnrlistC         = jjnr[jidx+2];
 360             jnrlistD         = jjnr[jidx+3];
 361             jnrlistE         = jjnr[jidx+4];
 362             jnrlistF         = jjnr[jidx+5];
 363             jnrlistG         = jjnr[jidx+6];
 364             jnrlistH         = jjnr[jidx+7];
 365             /* Sign of each element will be negative for non-real atoms.
 366              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 367              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 368              */
 369             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 370                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 371
 372             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 373             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 374             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 375             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 376             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 377             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 378             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 379             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 380             j_coord_offsetA  = DIM*jnrA;
 381             j_coord_offsetB  = DIM*jnrB;
 382             j_coord_offsetC  = DIM*jnrC;
 383             j_coord_offsetD  = DIM*jnrD;
 384             j_coord_offsetE  = DIM*jnrE;
 385             j_coord_offsetF  = DIM*jnrF;
 386             j_coord_offsetG  = DIM*jnrG;
 387             j_coord_offsetH  = DIM*jnrH;
 388
 389             /* load j atom coordinates */
 390             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 391                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 392                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 393                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 394                                                  &jx0,&jy0,&jz0);
 395
 396             /* Calculate displacement vector */
 397             dx00             = _mm256_sub_ps(ix0,jx0);
 398             dy00             = _mm256_sub_ps(iy0,jy0);
 399             dz00             = _mm256_sub_ps(iz0,jz0);
 400
 401             /* Calculate squared distance and things based on it */
 402             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 403
 404             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 405
 406             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 407
 408             /* Load parameters for j particles */
 409             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 410                                                                  charge+jnrC+0,charge+jnrD+0,
 411                                                                  charge+jnrE+0,charge+jnrF+0,
 412                                                                  charge+jnrG+0,charge+jnrH+0);
 413             vdwjidx0A        = 2*vdwtype[jnrA+0];
 414             vdwjidx0B        = 2*vdwtype[jnrB+0];
 415             vdwjidx0C        = 2*vdwtype[jnrC+0];
 416             vdwjidx0D        = 2*vdwtype[jnrD+0];
 417             vdwjidx0E        = 2*vdwtype[jnrE+0];
 418             vdwjidx0F        = 2*vdwtype[jnrF+0];
 419             vdwjidx0G        = 2*vdwtype[jnrG+0];
 420             vdwjidx0H        = 2*vdwtype[jnrH+0];
 421
 422             /**************************
 423              * CALCULATE INTERACTIONS *
 424              **************************/
 425
 426             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 427             {
 428
 429             r00              = _mm256_mul_ps(rsq00,rinv00);
 430             r00              = _mm256_andnot_ps(dummy_mask,r00);
 431
 432             /* Compute parameters for interactions between i and j atoms */
 433             qq00             = _mm256_mul_ps(iq0,jq0);
 434             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 435                                             vdwioffsetptr0+vdwjidx0B,
 436                                             vdwioffsetptr0+vdwjidx0C,
 437                                             vdwioffsetptr0+vdwjidx0D,
 438                                             vdwioffsetptr0+vdwjidx0E,
 439                                             vdwioffsetptr0+vdwjidx0F,
 440                                             vdwioffsetptr0+vdwjidx0G,
 441                                             vdwioffsetptr0+vdwjidx0H,
 442                                             &c6_00,&c12_00);
 443
 444             c6grid_00       = gmx_mm256_load_8real_swizzle_ps(vdwgridioffsetptr0+vdwjidx0A,
 445                                                                   vdwgridioffsetptr0+vdwjidx0B,
 446                                                                   vdwgridioffsetptr0+vdwjidx0C,
 447                                                                   vdwgridioffsetptr0+vdwjidx0D,
 448                                                                   vdwgridioffsetptr0+vdwjidx0E,
 449                                                                   vdwgridioffsetptr0+vdwjidx0F,
 450                                                                   vdwgridioffsetptr0+vdwjidx0G,
 451                                                                   vdwgridioffsetptr0+vdwjidx0H);
 452
 453             /* EWALD ELECTROSTATICS */
 454
 455             /* Analytical PME correction */
 456             zeta2            = _mm256_mul_ps(beta2,rsq00);
 457             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 458             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 459             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 460             felec            = _mm256_mul_ps(qq00,felec);
 461             pmecorrV         = gmx_mm256_pmecorrV_ps(zeta2);
 462             pmecorrV         = _mm256_mul_ps(pmecorrV,beta);
 463             velec            = _mm256_sub_ps(_mm256_sub_ps(rinv00,sh_ewald),pmecorrV);
 464             velec            = _mm256_mul_ps(qq00,velec);
 465
 466             /* Analytical LJ-PME */
 467             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 468             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 469             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 470             exponent         = gmx_simd_exp_r(ewcljrsq);
 471             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 472             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 473             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 474             vvdw6            = _mm256_mul_ps(_mm256_sub_ps(c6_00,_mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly))),rinvsix);
 475             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 476             vvdw             = _mm256_sub_ps(_mm256_mul_ps( _mm256_sub_ps(vvdw12 , _mm256_mul_ps(c12_00,_mm256_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
 477                                           _mm256_mul_ps( _mm256_sub_ps(vvdw6,_mm256_add_ps(_mm256_mul_ps(c6_00,sh_vdw_invrcut6),_mm256_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
 478             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 479             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,_mm256_sub_ps(vvdw6,_mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6)))),rinvsq00);
 480
 481             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 482
 483             /* Update potential sum for this i atom from the interaction with this j atom. */
 484             velec            = _mm256_and_ps(velec,cutoff_mask);
 485             velec            = _mm256_andnot_ps(dummy_mask,velec);
 486             velecsum         = _mm256_add_ps(velecsum,velec);
 487             vvdw             = _mm256_and_ps(vvdw,cutoff_mask);
 488             vvdw             = _mm256_andnot_ps(dummy_mask,vvdw);
 489             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 490
 491             fscal            = _mm256_add_ps(felec,fvdw);
 492
 493             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 494
 495             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 496
 497             /* Calculate temporary vectorial force */
 498             tx               = _mm256_mul_ps(fscal,dx00);
 499             ty               = _mm256_mul_ps(fscal,dy00);
 500             tz               = _mm256_mul_ps(fscal,dz00);
 501
 502             /* Update vectorial force */
 503             fix0             = _mm256_add_ps(fix0,tx);
 504             fiy0             = _mm256_add_ps(fiy0,ty);
 505             fiz0             = _mm256_add_ps(fiz0,tz);
 506
 507             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 508             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 509             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 510             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 511             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
 512             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
 513             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
 514             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
 515             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,tx,ty,tz);
 516
 517             }
 518
 519             /* Inner loop uses 146 flops */
 520         }
 521
 522         /* End of innermost loop */
 523
 524         gmx_mm256_update_iforce_1atom_swizzle_ps(fix0,fiy0,fiz0,
 525                                                  f+i_coord_offset,fshift+i_shift_offset);
 526
 527         ggid                        = gid[iidx];
 528         /* Update potential energies */
 529         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 530         gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 531
 532         /* Increment number of inner iterations */
 533         inneriter                  += j_index_end - j_index_start;
 534
 535         /* Outer loop uses 9 flops */
 536     }
 537
 538     /* Increment number of outer iterations */
 539     outeriter        += nri;
 540
 541     /* Update outer/inner flops */
 542
 543     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*146);
 544 }
 545 /*
 546  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_F_avx_256_single
 547  * Electrostatics interaction: Ewald
 548  * VdW interaction:            LJEwald
 549  * Geometry:                   Particle-Particle
 550  * Calculate force/pot:        Force
 551  */
 552 void
 553 nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_F_avx_256_single
 554                     (t_nblist                    * gmx_restrict       nlist,
 555                      rvec                        * gmx_restrict          xx,
 556                      rvec                        * gmx_restrict          ff,
 557                      t_forcerec                  * gmx_restrict          fr,
 558                      t_mdatoms                   * gmx_restrict     mdatoms,
 559                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 560                      t_nrnb                      * gmx_restrict        nrnb)
 561 {
 562     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 563      * just 0 for non-waters.
 564      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
 565      * jnr indices corresponding to data put in the four positions in the SIMD register.
 566      */
 567     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 568     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 569     int              jnrA,jnrB,jnrC,jnrD;
 570     int              jnrE,jnrF,jnrG,jnrH;
 571     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 572     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 573     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 574     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
 575     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 576     real             rcutoff_scalar;
 577     real             *shiftvec,*fshift,*x,*f;
 578     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
 579     real             scratch[4*DIM];
 580     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 581     real *           vdwioffsetptr0;
 582     real *           vdwgridioffsetptr0;
 583     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 584     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
 585     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 586     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 587     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
 588     real             *charge;
 589     int              nvdwtype;
 590     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 591     int              *vdwtype;
 592     real             *vdwparam;
 593     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 594     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 595     __m256           c6grid_00;
 596     real             *vdwgridparam;
 597     __m256           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 598     __m256           one_half  = _mm256_set1_ps(0.5);
 599     __m256           minus_one = _mm256_set1_ps(-1.0);
 600     __m256i          ewitab;
 601     __m128i          ewitab_lo,ewitab_hi;
 602     __m256           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 603     __m256           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 604     real             *ewtab;
 605     __m256           dummy_mask,cutoff_mask;
 606     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 607     __m256           one     = _mm256_set1_ps(1.0);
 608     __m256           two     = _mm256_set1_ps(2.0);
 609     x                = xx[0];
 610     f                = ff[0];
 611
 612     nri              = nlist->nri;
 613     iinr             = nlist->iinr;
 614     jindex           = nlist->jindex;
 615     jjnr             = nlist->jjnr;
 616     shiftidx         = nlist->shift;
 617     gid              = nlist->gid;
 618     shiftvec         = fr->shift_vec[0];
 619     fshift           = fr->fshift[0];
 620     facel            = _mm256_set1_ps(fr->epsfac);
 621     charge           = mdatoms->chargeA;
 622     nvdwtype         = fr->ntype;
 623     vdwparam         = fr->nbfp;
 624     vdwtype          = mdatoms->typeA;
 625     vdwgridparam     = fr->ljpme_c6grid;
 626     sh_lj_ewald      = _mm256_set1_ps(fr->ic->sh_lj_ewald);
 627     ewclj            = _mm256_set1_ps(fr->ewaldcoeff_lj);
 628     ewclj2           = _mm256_mul_ps(minus_one,_mm256_mul_ps(ewclj,ewclj));
 629
 630     sh_ewald         = _mm256_set1_ps(fr->ic->sh_ewald);
 631     beta             = _mm256_set1_ps(fr->ic->ewaldcoeff_q);
 632     beta2            = _mm256_mul_ps(beta,beta);
 633     beta3            = _mm256_mul_ps(beta,beta2);
 634
 635     ewtab            = fr->ic->tabq_coul_F;
 636     ewtabscale       = _mm256_set1_ps(fr->ic->tabq_scale);
 637     ewtabhalfspace   = _mm256_set1_ps(0.5/fr->ic->tabq_scale);
 638
 639     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 640     rcutoff_scalar   = fr->rcoulomb;
 641     rcutoff          = _mm256_set1_ps(rcutoff_scalar);
 642     rcutoff2         = _mm256_mul_ps(rcutoff,rcutoff);
 643
 644     sh_vdw_invrcut6  = _mm256_set1_ps(fr->ic->sh_invrc6);
 645     rvdw             = _mm256_set1_ps(fr->rvdw);
 646
 647     /* Avoid stupid compiler warnings */
 648     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 649     j_coord_offsetA = 0;
 650     j_coord_offsetB = 0;
 651     j_coord_offsetC = 0;
 652     j_coord_offsetD = 0;
 653     j_coord_offsetE = 0;
 654     j_coord_offsetF = 0;
 655     j_coord_offsetG = 0;
 656     j_coord_offsetH = 0;
 657
 658     outeriter        = 0;
 659     inneriter        = 0;
 660
 661     for(iidx=0;iidx<4*DIM;iidx++)
 662     {
 663         scratch[iidx] = 0.0;
 664     }
 665
 666     /* Start outer loop over neighborlists */
 667     for(iidx=0; iidx<nri; iidx++)
 668     {
 669         /* Load shift vector for this list */
 670         i_shift_offset   = DIM*shiftidx[iidx];
 671
 672         /* Load limits for loop over neighbors */
 673         j_index_start    = jindex[iidx];
 674         j_index_end      = jindex[iidx+1];
 675
 676         /* Get outer coordinate index */
 677         inr              = iinr[iidx];
 678         i_coord_offset   = DIM*inr;
 679
 680         /* Load i particle coords and add shift vector */
 681         gmx_mm256_load_shift_and_1rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 682
 683         fix0             = _mm256_setzero_ps();
 684         fiy0             = _mm256_setzero_ps();
 685         fiz0             = _mm256_setzero_ps();
 686
 687         /* Load parameters for i particles */
 688         iq0              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+0]));
 689         vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 690         vdwgridioffsetptr0 = vdwgridparam+2*nvdwtype*vdwtype[inr+0];
 691
 692         /* Start inner kernel loop */
 693         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 694         {
 695
 696             /* Get j neighbor index, and coordinate index */
 697             jnrA             = jjnr[jidx];
 698             jnrB             = jjnr[jidx+1];
 699             jnrC             = jjnr[jidx+2];
 700             jnrD             = jjnr[jidx+3];
 701             jnrE             = jjnr[jidx+4];
 702             jnrF             = jjnr[jidx+5];
 703             jnrG             = jjnr[jidx+6];
 704             jnrH             = jjnr[jidx+7];
 705             j_coord_offsetA  = DIM*jnrA;
 706             j_coord_offsetB  = DIM*jnrB;
 707             j_coord_offsetC  = DIM*jnrC;
 708             j_coord_offsetD  = DIM*jnrD;
 709             j_coord_offsetE  = DIM*jnrE;
 710             j_coord_offsetF  = DIM*jnrF;
 711             j_coord_offsetG  = DIM*jnrG;
 712             j_coord_offsetH  = DIM*jnrH;
 713
 714             /* load j atom coordinates */
 715             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 716                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 717                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 718                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 719                                                  &jx0,&jy0,&jz0);
 720
 721             /* Calculate displacement vector */
 722             dx00             = _mm256_sub_ps(ix0,jx0);
 723             dy00             = _mm256_sub_ps(iy0,jy0);
 724             dz00             = _mm256_sub_ps(iz0,jz0);
 725
 726             /* Calculate squared distance and things based on it */
 727             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 728
 729             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 730
 731             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 732
 733             /* Load parameters for j particles */
 734             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 735                                                                  charge+jnrC+0,charge+jnrD+0,
 736                                                                  charge+jnrE+0,charge+jnrF+0,
 737                                                                  charge+jnrG+0,charge+jnrH+0);
 738             vdwjidx0A        = 2*vdwtype[jnrA+0];
 739             vdwjidx0B        = 2*vdwtype[jnrB+0];
 740             vdwjidx0C        = 2*vdwtype[jnrC+0];
 741             vdwjidx0D        = 2*vdwtype[jnrD+0];
 742             vdwjidx0E        = 2*vdwtype[jnrE+0];
 743             vdwjidx0F        = 2*vdwtype[jnrF+0];
 744             vdwjidx0G        = 2*vdwtype[jnrG+0];
 745             vdwjidx0H        = 2*vdwtype[jnrH+0];
 746
 747             /**************************
 748              * CALCULATE INTERACTIONS *
 749              **************************/
 750
 751             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 752             {
 753
 754             r00              = _mm256_mul_ps(rsq00,rinv00);
 755
 756             /* Compute parameters for interactions between i and j atoms */
 757             qq00             = _mm256_mul_ps(iq0,jq0);
 758             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 759                                             vdwioffsetptr0+vdwjidx0B,
 760                                             vdwioffsetptr0+vdwjidx0C,
 761                                             vdwioffsetptr0+vdwjidx0D,
 762                                             vdwioffsetptr0+vdwjidx0E,
 763                                             vdwioffsetptr0+vdwjidx0F,
 764                                             vdwioffsetptr0+vdwjidx0G,
 765                                             vdwioffsetptr0+vdwjidx0H,
 766                                             &c6_00,&c12_00);
 767
 768             c6grid_00       = gmx_mm256_load_8real_swizzle_ps(vdwgridioffsetptr0+vdwjidx0A,
 769                                                                   vdwgridioffsetptr0+vdwjidx0B,
 770                                                                   vdwgridioffsetptr0+vdwjidx0C,
 771                                                                   vdwgridioffsetptr0+vdwjidx0D,
 772                                                                   vdwgridioffsetptr0+vdwjidx0E,
 773                                                                   vdwgridioffsetptr0+vdwjidx0F,
 774                                                                   vdwgridioffsetptr0+vdwjidx0G,
 775                                                                   vdwgridioffsetptr0+vdwjidx0H);
 776
 777             /* EWALD ELECTROSTATICS */
 778
 779             /* Analytical PME correction */
 780             zeta2            = _mm256_mul_ps(beta2,rsq00);
 781             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 782             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 783             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 784             felec            = _mm256_mul_ps(qq00,felec);
 785
 786             /* Analytical LJ-PME */
 787             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 788             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 789             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 790             exponent         = gmx_simd_exp_r(ewcljrsq);
 791             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 792             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 793             /* f6A = 6 * C6grid * (1 - poly) */
 794             f6A              = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
 795             /* f6B = C6grid * exponent * beta^6 */
 796             f6B              = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
 797             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 798             fvdw              = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
 799
 800             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 801
 802             fscal            = _mm256_add_ps(felec,fvdw);
 803
 804             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 805
 806             /* Calculate temporary vectorial force */
 807             tx               = _mm256_mul_ps(fscal,dx00);
 808             ty               = _mm256_mul_ps(fscal,dy00);
 809             tz               = _mm256_mul_ps(fscal,dz00);
 810
 811             /* Update vectorial force */
 812             fix0             = _mm256_add_ps(fix0,tx);
 813             fiy0             = _mm256_add_ps(fiy0,ty);
 814             fiz0             = _mm256_add_ps(fiz0,tz);
 815
 816             fjptrA             = f+j_coord_offsetA;
 817             fjptrB             = f+j_coord_offsetB;
 818             fjptrC             = f+j_coord_offsetC;
 819             fjptrD             = f+j_coord_offsetD;
 820             fjptrE             = f+j_coord_offsetE;
 821             fjptrF             = f+j_coord_offsetF;
 822             fjptrG             = f+j_coord_offsetG;
 823             fjptrH             = f+j_coord_offsetH;
 824             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,tx,ty,tz);
 825
 826             }
 827
 828             /* Inner loop uses 82 flops */
 829         }
 830
 831         if(jidx<j_index_end)
 832         {
 833
 834             /* Get j neighbor index, and coordinate index */
 835             jnrlistA         = jjnr[jidx];
 836             jnrlistB         = jjnr[jidx+1];
 837             jnrlistC         = jjnr[jidx+2];
 838             jnrlistD         = jjnr[jidx+3];
 839             jnrlistE         = jjnr[jidx+4];
 840             jnrlistF         = jjnr[jidx+5];
 841             jnrlistG         = jjnr[jidx+6];
 842             jnrlistH         = jjnr[jidx+7];
 843             /* Sign of each element will be negative for non-real atoms.
 844              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 845              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 846              */
 847             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 848                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 849
 850             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 851             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 852             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 853             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 854             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 855             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 856             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 857             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 858             j_coord_offsetA  = DIM*jnrA;
 859             j_coord_offsetB  = DIM*jnrB;
 860             j_coord_offsetC  = DIM*jnrC;
 861             j_coord_offsetD  = DIM*jnrD;
 862             j_coord_offsetE  = DIM*jnrE;
 863             j_coord_offsetF  = DIM*jnrF;
 864             j_coord_offsetG  = DIM*jnrG;
 865             j_coord_offsetH  = DIM*jnrH;
 866
 867             /* load j atom coordinates */
 868             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 869                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 870                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 871                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 872                                                  &jx0,&jy0,&jz0);
 873
 874             /* Calculate displacement vector */
 875             dx00             = _mm256_sub_ps(ix0,jx0);
 876             dy00             = _mm256_sub_ps(iy0,jy0);
 877             dz00             = _mm256_sub_ps(iz0,jz0);
 878
 879             /* Calculate squared distance and things based on it */
 880             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 881
 882             rinv00           = gmx_mm256_invsqrt_ps(rsq00);
 883
 884             rinvsq00         = _mm256_mul_ps(rinv00,rinv00);
 885
 886             /* Load parameters for j particles */
 887             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 888                                                                  charge+jnrC+0,charge+jnrD+0,
 889                                                                  charge+jnrE+0,charge+jnrF+0,
 890                                                                  charge+jnrG+0,charge+jnrH+0);
 891             vdwjidx0A        = 2*vdwtype[jnrA+0];
 892             vdwjidx0B        = 2*vdwtype[jnrB+0];
 893             vdwjidx0C        = 2*vdwtype[jnrC+0];
 894             vdwjidx0D        = 2*vdwtype[jnrD+0];
 895             vdwjidx0E        = 2*vdwtype[jnrE+0];
 896             vdwjidx0F        = 2*vdwtype[jnrF+0];
 897             vdwjidx0G        = 2*vdwtype[jnrG+0];
 898             vdwjidx0H        = 2*vdwtype[jnrH+0];
 899
 900             /**************************
 901              * CALCULATE INTERACTIONS *
 902              **************************/
 903
 904             if (gmx_mm256_any_lt(rsq00,rcutoff2))
 905             {
 906
 907             r00              = _mm256_mul_ps(rsq00,rinv00);
 908             r00              = _mm256_andnot_ps(dummy_mask,r00);
 909
 910             /* Compute parameters for interactions between i and j atoms */
 911             qq00             = _mm256_mul_ps(iq0,jq0);
 912             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 913                                             vdwioffsetptr0+vdwjidx0B,
 914                                             vdwioffsetptr0+vdwjidx0C,
 915                                             vdwioffsetptr0+vdwjidx0D,
 916                                             vdwioffsetptr0+vdwjidx0E,
 917                                             vdwioffsetptr0+vdwjidx0F,
 918                                             vdwioffsetptr0+vdwjidx0G,
 919                                             vdwioffsetptr0+vdwjidx0H,
 920                                             &c6_00,&c12_00);
 921
 922             c6grid_00       = gmx_mm256_load_8real_swizzle_ps(vdwgridioffsetptr0+vdwjidx0A,
 923                                                                   vdwgridioffsetptr0+vdwjidx0B,
 924                                                                   vdwgridioffsetptr0+vdwjidx0C,
 925                                                                   vdwgridioffsetptr0+vdwjidx0D,
 926                                                                   vdwgridioffsetptr0+vdwjidx0E,
 927                                                                   vdwgridioffsetptr0+vdwjidx0F,
 928                                                                   vdwgridioffsetptr0+vdwjidx0G,
 929                                                                   vdwgridioffsetptr0+vdwjidx0H);
 930
 931             /* EWALD ELECTROSTATICS */
 932
 933             /* Analytical PME correction */
 934             zeta2            = _mm256_mul_ps(beta2,rsq00);
 935             rinv3            = _mm256_mul_ps(rinvsq00,rinv00);
 936             pmecorrF         = gmx_mm256_pmecorrF_ps(zeta2);
 937             felec            = _mm256_add_ps( _mm256_mul_ps(pmecorrF,beta3), rinv3);
 938             felec            = _mm256_mul_ps(qq00,felec);
 939
 940             /* Analytical LJ-PME */
 941             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 942             ewcljrsq         = _mm256_mul_ps(ewclj2,rsq00);
 943             ewclj6           = _mm256_mul_ps(ewclj2,_mm256_mul_ps(ewclj2,ewclj2));
 944             exponent         = gmx_simd_exp_r(ewcljrsq);
 945             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 946             poly             = _mm256_mul_ps(exponent,_mm256_add_ps(_mm256_sub_ps(one,ewcljrsq),_mm256_mul_ps(_mm256_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 947             /* f6A = 6 * C6grid * (1 - poly) */
 948             f6A              = _mm256_mul_ps(c6grid_00,_mm256_sub_ps(one,poly));
 949             /* f6B = C6grid * exponent * beta^6 */
 950             f6B              = _mm256_mul_ps(_mm256_mul_ps(c6grid_00,one_sixth),_mm256_mul_ps(exponent,ewclj6));
 951             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 952             fvdw              = _mm256_mul_ps(_mm256_add_ps(_mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),_mm256_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
 953
 954             cutoff_mask      = _mm256_cmp_ps(rsq00,rcutoff2,_CMP_LT_OQ);
 955
 956             fscal            = _mm256_add_ps(felec,fvdw);
 957
 958             fscal            = _mm256_and_ps(fscal,cutoff_mask);
 959
 960             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 961
 962             /* Calculate temporary vectorial force */
 963             tx               = _mm256_mul_ps(fscal,dx00);
 964             ty               = _mm256_mul_ps(fscal,dy00);
 965             tz               = _mm256_mul_ps(fscal,dz00);
 966
 967             /* Update vectorial force */
 968             fix0             = _mm256_add_ps(fix0,tx);
 969             fiy0             = _mm256_add_ps(fiy0,ty);
 970             fiz0             = _mm256_add_ps(fiz0,tz);
 971
 972             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 973             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 974             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 975             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 976             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
 977             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
 978             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
 979             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
 980             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,tx,ty,tz);
 981
 982             }
 983
 984             /* Inner loop uses 83 flops */
 985         }
 986
 987         /* End of innermost loop */
 988
 989         gmx_mm256_update_iforce_1atom_swizzle_ps(fix0,fiy0,fiz0,
 990                                                  f+i_coord_offset,fshift+i_shift_offset);
 991
 992         /* Increment number of inner iterations */
 993         inneriter                  += j_index_end - j_index_start;
 994
 995         /* Outer loop uses 7 flops */
 996     }
 997
 998     /* Increment number of outer iterations */
 999     outeriter        += nri;
1000
1001     /* Update outer/inner flops */
1002
1003     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*83);
1004 }