src/gromacs/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_template_avx_256_double.pre

   1 /* #if 0 */
   2 /*
   3  * This file is part of the GROMACS molecular simulation package.
   4  *
   5  * Copyright (c) 2012,2013,2014,2015,2017, by the GROMACS development team, led by
   6  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   7  * and including many others, as listed in the AUTHORS file in the
   8  * top-level source directory and at http://www.gromacs.org.
   9  *
  10  * GROMACS is free software; you can redistribute it and/or
  11  * modify it under the terms of the GNU Lesser General Public License
  12  * as published by the Free Software Foundation; either version 2.1
  13  * of the License, or (at your option) any later version.
  14  *
  15  * GROMACS is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18  * Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public
  21  * License along with GROMACS; if not, see
  22  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  23  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  24  *
  25  * If you want to redistribute modifications to GROMACS, please
  26  * consider that scientific software is very special. Version
  27  * control is crucial - bugs must be traceable. We will be happy to
  28  * consider code for inclusion in the official distribution, but
  29  * derived work must not be called official GROMACS. Details are found
  30  * in the README & COPYING files - if they are missing, get the
  31  * official version at http://www.gromacs.org.
  32  *
  33  * To help us fund GROMACS development, we humbly ask that you cite
  34  * the research papers on the package. Check out http://www.gromacs.org.
  35  */
  36 #error This file must be processed with the Gromacs pre-preprocessor
  37 /* #endif */
  38 /* #if INCLUDE_HEADER */
  39 #include "gmxpre.h"
  40
  41 #include "config.h"
  42
  43 #include <math.h>
  44
  45 #include "../nb_kernel.h"
  46 #include "gromacs/gmxlib/nrnb.h"
  47
  48 #include "kernelutil_x86_avx_256_double.h"
  49 /* #endif */
  50
  51 /* ## List of variables set by the generating script:                                    */
  52 /* ##                                                                                    */
  53 /* ## Setttings that apply to the entire kernel:                                         */
  54 /* ## KERNEL_ELEC:           String, choice for electrostatic interactions               */
  55 /* ## KERNEL_VDW:            String, choice for van der Waals interactions               */
  56 /* ## KERNEL_NAME:           String, name of this kernel                                 */
  57 /* ## KERNEL_VF:             String telling if we calculate potential, force, or both    */
  58 /* ## GEOMETRY_I/GEOMETRY_J: String, name of each geometry, e.g. 'Water3' or '1Particle' */
  59 /* ##                                                                                    */
  60 /* ## Setttings that apply to particles in the outer (I) or inner (J) loops:             */
  61 /* ## PARTICLES_I[]/         Arrays with lists of i/j particles to use in kernel. It is  */
  62 /* ## PARTICLES_J[]:         just [0] for particle geometry, but can be longer for water */
  63 /* ## PARTICLES_ELEC_I[]/    Arrays with lists of i/j particle that have electrostatics  */
  64 /* ## PARTICLES_ELEC_J[]:    interactions that should be calculated in this kernel.      */
  65 /* ## PARTICLES_VDW_I[]/     Arrays with the list of i/j particle that have VdW          */
  66 /* ## PARTICLES_VDW_J[]:     interactions that should be calculated in this kernel.      */
  67 /* ##                                                                                    */
  68 /* ## Setttings for pairs of interactions (e.g. 2nd i particle against 1st j particle)   */
  69 /* ## PAIRS_IJ[]:            Array with (i,j) tuples of pairs for which interactions     */
  70 /* ##                        should be calculated in this kernel. Zero-charge particles  */
  71 /* ##                        do not have interactions with particles without vdw, and    */
  72 /* ##                        Vdw-only interactions are not evaluated in a no-vdw-kernel. */
  73 /* ## INTERACTION_FLAGS[][]: 2D matrix, dimension e.g. 3*3 for water-water interactions. */
  74 /* ##                        For each i-j pair, the element [I][J] is a list of strings  */
  75 /* ##                        defining properties/flags of this interaction. Examples     */
  76 /* ##                        include 'electrostatics'/'vdw' if that type of interaction  */
  77 /* ##                        should be evaluated, 'rsq'/'rinv'/'rinvsq' if those values  */
  78 /* ##                        are needed, and 'exactcutoff' or 'shift','switch' to        */
  79 /* ##                        decide if the force/potential should be modified. This way  */
  80 /* ##                        we only calculate values absolutely needed for each case.   */
  81
  82 /* ## Calculate the size and offset for (merged/interleaved) table data */
  83
  84 /*
  85  * Gromacs nonbonded kernel:   {KERNEL_NAME}
  86  * Electrostatics interaction: {KERNEL_ELEC}
  87  * VdW interaction:            {KERNEL_VDW}
  88  * Geometry:                   {GEOMETRY_I}-{GEOMETRY_J}
  89  * Calculate force/pot:        {KERNEL_VF}
  90  */
  91 void
  92 {KERNEL_NAME}
  93                     (t_nblist                    * gmx_restrict       nlist,
  94                      rvec                        * gmx_restrict          xx,
  95                      rvec                        * gmx_restrict          ff,
  96                      struct t_forcerec           * gmx_restrict          fr,
  97                      t_mdatoms                   * gmx_restrict     mdatoms,
  98                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  99                      t_nrnb                      * gmx_restrict        nrnb)
 100 {
 101     /* ## Not all variables are used for all kernels, but any optimizing compiler fixes that, */
 102     /* ## so there is no point in going to extremes to exclude variables that are not needed. */
 103     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 104      * just 0 for non-waters.
 105      * Suffixes A,B,C,D refer to j loop unrolling done with AVX, e.g. for the four different
 106      * jnr indices corresponding to data put in the four positions in the SIMD register.
 107      */
 108     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 109     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 110     int              jnrA,jnrB,jnrC,jnrD;
 111     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 112     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 113     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 114     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 115     real             rcutoff_scalar;
 116     real             *shiftvec,*fshift,*x,*f;
 117     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 118     real             scratch[4*DIM];
 119     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 120     /* #for I in PARTICLES_I */
 121     real *           vdwioffsetptr{I};
 122     /* #if 'LJEwald' in KERNEL_VDW */
 123     real *           vdwgridioffsetptr{I};
 124     /* #endif                      */
 125     __m256d          ix{I},iy{I},iz{I},fix{I},fiy{I},fiz{I},iq{I},isai{I};
 126     /* #endfor */
 127     /* #for J in PARTICLES_J */
 128     int              vdwjidx{J}A,vdwjidx{J}B,vdwjidx{J}C,vdwjidx{J}D;
 129     __m256d          jx{J},jy{J},jz{J},fjx{J},fjy{J},fjz{J},jq{J},isaj{J};
 130     /* #endfor */
 131     /* #for I,J in PAIRS_IJ */
 132     __m256d          dx{I}{J},dy{I}{J},dz{I}{J},rsq{I}{J},rinv{I}{J},rinvsq{I}{J},r{I}{J},qq{I}{J},c6_{I}{J},c12_{I}{J};
 133     /* #endfor */
 134     /* #if KERNEL_ELEC != 'None' */
 135     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
 136     real             *charge;
 137     /* #endif */
 138     /* #if 'GeneralizedBorn' in KERNEL_ELEC */
 139     __m128i          gbitab;
 140     __m256d          vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,gbeps,dvdatmp;
 141     __m256d          minushalf = _mm256_set1_pd(-0.5);
 142     real             *invsqrta,*dvda,*gbtab;
 143     /* #endif */
 144     /* #if KERNEL_VDW != 'None' */
 145     int              nvdwtype;
 146     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 147     int              *vdwtype;
 148     real             *vdwparam;
 149     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
 150     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
 151     /* #endif */
 152     /* #if 'Table' in KERNEL_ELEC or 'GeneralizedBorn' in KERNEL_ELEC or 'Table' in KERNEL_VDW */
 153     __m128i          vfitab;
 154     __m128i          ifour       = _mm_set1_epi32(4);
 155     __m256d          rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
 156     real             *vftab;
 157     /* #endif */
 158     /* #if 'LJEwald' in KERNEL_VDW */
 159     /* #for I,J in PAIRS_IJ */
 160     __m256d           c6grid_{I}{J};
 161     /* #endfor */
 162     real             *vdwgridparam;
 163     __m256d           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 164     __m256d           one_half  = _mm256_set1_pd(0.5);
 165     __m256d           minus_one = _mm256_set1_pd(-1.0);
 166     /* #endif */
 167     /* #if 'Ewald' in KERNEL_ELEC */
 168     __m128i          ewitab;
 169     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 170     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 171     real             *ewtab;
 172     /* #endif */
 173     /* #if 'PotentialSwitch' in [KERNEL_MOD_ELEC,KERNEL_MOD_VDW] */
 174     __m256d          rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 175     real             rswitch_scalar,d_scalar;
 176     /* #endif */
 177     __m256d          dummy_mask,cutoff_mask;
 178     __m128           tmpmask0,tmpmask1;
 179     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
 180     __m256d          one     = _mm256_set1_pd(1.0);
 181     __m256d          two     = _mm256_set1_pd(2.0);
 182     x                = xx[0];
 183     f                = ff[0];
 184
 185     nri              = nlist->nri;
 186     iinr             = nlist->iinr;
 187     jindex           = nlist->jindex;
 188     jjnr             = nlist->jjnr;
 189     shiftidx         = nlist->shift;
 190     gid              = nlist->gid;
 191     shiftvec         = fr->shift_vec[0];
 192     fshift           = fr->fshift[0];
 193     /* #if KERNEL_ELEC != 'None' */
 194     facel            = _mm256_set1_pd(fr->ic->epsfac);
 195     charge           = mdatoms->chargeA;
 196     /*     #if 'ReactionField' in KERNEL_ELEC */
 197     krf              = _mm256_set1_pd(fr->ic->k_rf);
 198     krf2             = _mm256_set1_pd(fr->ic->k_rf*2.0);
 199     crf              = _mm256_set1_pd(fr->ic->c_rf);
 200     /*     #endif */
 201     /* #endif */
 202     /* #if KERNEL_VDW != 'None' */
 203     nvdwtype         = fr->ntype;
 204     vdwparam         = fr->nbfp;
 205     vdwtype          = mdatoms->typeA;
 206     /* #endif */
 207     /* #if 'LJEwald' in KERNEL_VDW */
 208     vdwgridparam     = fr->ljpme_c6grid;
 209     sh_lj_ewald      = _mm256_set1_pd(fr->ic->sh_lj_ewald);
 210     ewclj            = _mm256_set1_pd(fr->ic->ewaldcoeff_lj);
 211     ewclj2           = _mm256_mul_pd(minus_one,_mm256_mul_pd(ewclj,ewclj));
 212     /* #endif */
 213
 214     /* #if 'Table' in KERNEL_ELEC and 'Table' in KERNEL_VDW */
 215     vftab            = kernel_data->table_elec_vdw->data;
 216     vftabscale       = _mm256_set1_pd(kernel_data->table_elec_vdw->scale);
 217     /* #elif 'Table' in KERNEL_ELEC */
 218     vftab            = kernel_data->table_elec->data;
 219     vftabscale       = _mm256_set1_pd(kernel_data->table_elec->scale);
 220     /* #elif 'Table' in KERNEL_VDW */
 221     vftab            = kernel_data->table_vdw->data;
 222     vftabscale       = _mm256_set1_pd(kernel_data->table_vdw->scale);
 223     /* #endif */
 224
 225     /* #if 'Ewald' in KERNEL_ELEC */
 226     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
 227     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
 228     beta2            = _mm256_mul_pd(beta,beta);
 229     beta3            = _mm256_mul_pd(beta,beta2);
 230
 231     /*     #if KERNEL_VF=='Force' and KERNEL_MOD_ELEC!='PotentialSwitch' */
 232     ewtab            = fr->ic->tabq_coul_F;
 233     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 234     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 235     /*     #else */
 236     ewtab            = fr->ic->tabq_coul_FDV0;
 237     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 238     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 239      /*     #endif */
 240     /* #endif */
 241
 242     /* #if KERNEL_ELEC=='GeneralizedBorn' */
 243     invsqrta         = fr->invsqrta;
 244     dvda             = fr->dvda;
 245     gbtabscale       = _mm256_set1_pd(fr->gbtab->scale);
 246     gbtab            = fr->gbtab->data;
 247     gbinvepsdiff     = _mm256_set1_pd((1.0/fr->ic->epsilon_r) - (1.0/fr->gb_epsilon_solvent));
 248     /* #endif */
 249
 250     /* #if 'Water' in GEOMETRY_I */
 251     /* Setup water-specific parameters */
 252     inr              = nlist->iinr[0];
 253     /*     #for I in PARTICLES_ELEC_I */
 254     iq{I}              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+{I}]));
 255     /*     #endfor */
 256     /*     #for I in PARTICLES_VDW_I */
 257     vdwioffsetptr{I}   = vdwparam+2*nvdwtype*vdwtype[inr+{I}];
 258     /*         #if 'LJEwald' in KERNEL_VDW */
 259     vdwgridioffsetptr{I} = vdwgridparam+2*nvdwtype*vdwtype[inr+{I}];
 260     /*         #endif */
 261     /*     #endfor */
 262     /* #endif */
 263
 264     /* #if 'Water' in GEOMETRY_J */
 265     /*     #for J in PARTICLES_ELEC_J */
 266     jq{J}              = _mm256_set1_pd(charge[inr+{J}]);
 267     /*     #endfor */
 268     /*     #for J in PARTICLES_VDW_J */
 269     vdwjidx{J}A        = 2*vdwtype[inr+{J}];
 270     /*     #endfor */
 271     /*     #for I,J in PAIRS_IJ */
 272     /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 273     qq{I}{J}             = _mm256_mul_pd(iq{I},jq{J});
 274     /*         #endif */
 275     /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 276     /*             #if 'LJEwald' in KERNEL_VDW */
 277     c6_{I}{J}            = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A]);
 278     c12_{I}{J}           = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A+1]);
 279     c6grid_{I}{J}        = _mm256_set1_pd(vdwgridioffsetptr{I}[vdwjidx{J}A]);
 280     /*             #else */
 281     c6_{I}{J}            = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A]);
 282     c12_{I}{J}           = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A+1]);
 283     /*             #endif */
 284     /*         #endif */
 285     /*     #endfor */
 286     /* #endif */
 287
 288     /* #if KERNEL_MOD_ELEC!='None' or KERNEL_MOD_VDW!='None' */
 289     /*     #if KERNEL_ELEC!='None' */
 290     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 291     rcutoff_scalar   = fr->ic->rcoulomb;
 292     /*     #else */
 293     rcutoff_scalar   = fr->ic->rvdw;
 294     /*     #endif */
 295     rcutoff          = _mm256_set1_pd(rcutoff_scalar);
 296     rcutoff2         = _mm256_mul_pd(rcutoff,rcutoff);
 297     /* #endif */
 298
 299     /* #if KERNEL_MOD_VDW=='PotentialShift' */
 300     sh_vdw_invrcut6  = _mm256_set1_pd(fr->ic->sh_invrc6);
 301     rvdw             = _mm256_set1_pd(fr->ic->rvdw);
 302     /* #endif */
 303
 304     /* #if 'PotentialSwitch' in [KERNEL_MOD_ELEC,KERNEL_MOD_VDW] */
 305     /*     #if KERNEL_MOD_ELEC=='PotentialSwitch'  */
 306     rswitch_scalar   = fr->ic->rcoulomb_switch;
 307     rswitch          = _mm256_set1_pd(rswitch_scalar);
 308     /*     #else */
 309     rswitch_scalar   = fr->ic->rvdw_switch;
 310     rswitch          = _mm256_set1_pd(rswitch_scalar);
 311     /*     #endif */
 312     /* Setup switch parameters */
 313     d_scalar         = rcutoff_scalar-rswitch_scalar;
 314     d                = _mm256_set1_pd(d_scalar);
 315     swV3             = _mm256_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
 316     swV4             = _mm256_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 317     swV5             = _mm256_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 318     /*     #if 'Force' in KERNEL_VF */
 319     swF2             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
 320     swF3             = _mm256_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 321     swF4             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 322     /*     #endif */
 323     /* #endif */
 324
 325     /* Avoid stupid compiler warnings */
 326     jnrA = jnrB = jnrC = jnrD = 0;
 327     j_coord_offsetA = 0;
 328     j_coord_offsetB = 0;
 329     j_coord_offsetC = 0;
 330     j_coord_offsetD = 0;
 331
 332     /* ## Keep track of the floating point operations we issue for reporting! */
 333     /* #define OUTERFLOPS 0 */
 334     outeriter        = 0;
 335     inneriter        = 0;
 336
 337     for(iidx=0;iidx<4*DIM;iidx++)
 338     {
 339         scratch[iidx] = 0.0;
 340     }
 341
 342     /* Start outer loop over neighborlists */
 343     for(iidx=0; iidx<nri; iidx++)
 344     {
 345         /* Load shift vector for this list */
 346         i_shift_offset   = DIM*shiftidx[iidx];
 347
 348         /* Load limits for loop over neighbors */
 349         j_index_start    = jindex[iidx];
 350         j_index_end      = jindex[iidx+1];
 351
 352         /* Get outer coordinate index */
 353         inr              = iinr[iidx];
 354         i_coord_offset   = DIM*inr;
 355
 356         /* Load i particle coords and add shift vector */
 357         /* #if GEOMETRY_I == 'Particle' */
 358         gmx_mm256_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 359         /* #elif GEOMETRY_I == 'Water3' */
 360         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 361                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 362         /* #elif GEOMETRY_I == 'Water4' */
 363         /*     #if 0 in PARTICLES_I                 */
 364         gmx_mm256_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 365                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 366         /*     #else                                */
 367         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 368                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 369         /*     #endif                               */
 370         /* #endif                                   */
 371
 372         /* #if 'Force' in KERNEL_VF */
 373         /*     #for I in PARTICLES_I */
 374         fix{I}             = _mm256_setzero_pd();
 375         fiy{I}             = _mm256_setzero_pd();
 376         fiz{I}             = _mm256_setzero_pd();
 377         /*     #endfor */
 378         /* #endif */
 379
 380         /* ## For water we already preloaded parameters at the start of the kernel */
 381         /* #if not 'Water' in GEOMETRY_I */
 382         /* Load parameters for i particles */
 383         /*     #for I in PARTICLES_ELEC_I */
 384         iq{I}              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+{I}]));
 385         /*         #define OUTERFLOPS OUTERFLOPS+1 */
 386         /*         #if KERNEL_ELEC=='GeneralizedBorn' */
 387         isai{I}            = _mm256_set1_pd(invsqrta[inr+{I}]);
 388         /*         #endif */
 389         /*     #endfor */
 390         /*     #for I in PARTICLES_VDW_I */
 391         vdwioffsetptr{I}   = vdwparam+2*nvdwtype*vdwtype[inr+{I}];
 392         /*         #if 'LJEwald' in KERNEL_VDW */
 393         vdwgridioffsetptr{I} = vdwgridparam+2*nvdwtype*vdwtype[inr+{I}];
 394         /*         #endif */
 395         /*     #endfor */
 396         /* #endif */
 397
 398         /* #if 'Potential' in KERNEL_VF */
 399         /* Reset potential sums */
 400         /*     #if KERNEL_ELEC != 'None' */
 401         velecsum         = _mm256_setzero_pd();
 402         /*     #endif */
 403         /*     #if 'GeneralizedBorn' in KERNEL_ELEC */
 404         vgbsum           = _mm256_setzero_pd();
 405         /*     #endif */
 406         /*     #if KERNEL_VDW != 'None' */
 407         vvdwsum          = _mm256_setzero_pd();
 408         /*     #endif */
 409         /* #endif */
 410         /*     #if 'GeneralizedBorn' in KERNEL_ELEC and 'Force' in KERNEL_VF */
 411         dvdasum          = _mm256_setzero_pd();
 412         /*     #endif */
 413
 414         /* #for ROUND in ['Loop','Epilogue'] */
 415
 416         /* #if ROUND =='Loop' */
 417         /* Start inner kernel loop */
 418         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 419         {
 420         /* ## First round is normal loop (next statement resets indentation) */
 421         /*     #if 0 */
 422         }
 423         /*     #endif */
 424         /* #else */
 425         if(jidx<j_index_end)
 426         {
 427         /* ## Second round is epilogue */
 428         /* #endif */
 429         /* #define INNERFLOPS 0 */
 430
 431             /* Get j neighbor index, and coordinate index */
 432             /* #if ROUND =='Loop' */
 433             jnrA             = jjnr[jidx];
 434             jnrB             = jjnr[jidx+1];
 435             jnrC             = jjnr[jidx+2];
 436             jnrD             = jjnr[jidx+3];
 437             /* #else */
 438             jnrlistA         = jjnr[jidx];
 439             jnrlistB         = jjnr[jidx+1];
 440             jnrlistC         = jjnr[jidx+2];
 441             jnrlistD         = jjnr[jidx+3];
 442             /* Sign of each element will be negative for non-real atoms.
 443              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 444              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
 445              */
 446             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 447
 448             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
 449             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
 450             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
 451
 452             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 453             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 454             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 455             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 456             /* #endif */
 457             j_coord_offsetA  = DIM*jnrA;
 458             j_coord_offsetB  = DIM*jnrB;
 459             j_coord_offsetC  = DIM*jnrC;
 460             j_coord_offsetD  = DIM*jnrD;
 461
 462             /* load j atom coordinates */
 463             /* #if GEOMETRY_J == 'Particle'             */
 464             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 465                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 466                                                  &jx0,&jy0,&jz0);
 467             /* #elif GEOMETRY_J == 'Water3'             */
 468             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 469                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 470                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 471             /* #elif GEOMETRY_J == 'Water4'             */
 472             /*     #if 0 in PARTICLES_J                 */
 473             gmx_mm256_load_4rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 474                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 475                                                  &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
 476                                                  &jy2,&jz2,&jx3,&jy3,&jz3);
 477             /*     #else                                */
 478             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 479                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 480                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 481             /*     #endif                               */
 482             /* #endif                                   */
 483
 484             /* Calculate displacement vector */
 485             /* #for I,J in PAIRS_IJ */
 486             dx{I}{J}             = _mm256_sub_pd(ix{I},jx{J});
 487             dy{I}{J}             = _mm256_sub_pd(iy{I},jy{J});
 488             dz{I}{J}             = _mm256_sub_pd(iz{I},jz{J});
 489             /*     #define INNERFLOPS INNERFLOPS+3 */
 490             /* #endfor */
 491
 492             /* Calculate squared distance and things based on it */
 493             /* #for I,J in PAIRS_IJ */
 494             rsq{I}{J}            = gmx_mm256_calc_rsq_pd(dx{I}{J},dy{I}{J},dz{I}{J});
 495             /*     #define INNERFLOPS INNERFLOPS+5 */
 496             /* #endfor */
 497
 498             /* #for I,J in PAIRS_IJ */
 499             /*     #if 'rinv' in INTERACTION_FLAGS[I][J] */
 500             rinv{I}{J}           = avx256_invsqrt_d(rsq{I}{J});
 501             /*         #define INNERFLOPS INNERFLOPS+5 */
 502             /*     #endif */
 503             /* #endfor */
 504
 505             /* #for I,J in PAIRS_IJ */
 506             /*     #if 'rinvsq' in INTERACTION_FLAGS[I][J] */
 507             /*         # if 'rinv' not in INTERACTION_FLAGS[I][J] */
 508             rinvsq{I}{J}         = avx256_inv_d(rsq{I}{J});
 509             /*             #define INNERFLOPS INNERFLOPS+4 */
 510             /*         #else */
 511             rinvsq{I}{J}         = _mm256_mul_pd(rinv{I}{J},rinv{I}{J});
 512             /*             #define INNERFLOPS INNERFLOPS+1 */
 513             /*         #endif */
 514             /*     #endif */
 515             /* #endfor */
 516
 517             /* #if not 'Water' in GEOMETRY_J */
 518             /* Load parameters for j particles */
 519             /*     #for J in PARTICLES_ELEC_J */
 520             jq{J}              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+{J},charge+jnrB+{J},
 521                                                                  charge+jnrC+{J},charge+jnrD+{J});
 522             /*         #if KERNEL_ELEC=='GeneralizedBorn' */
 523             isaj{J}            = gmx_mm256_load_4real_swizzle_pd(invsqrta+jnrA+{J},invsqrta+jnrB+{J},
 524                                                                  invsqrta+jnrC+{J},invsqrta+jnrD+{J});
 525             /*         #endif */
 526             /*     #endfor */
 527             /*     #for J in PARTICLES_VDW_J */
 528             vdwjidx{J}A        = 2*vdwtype[jnrA+{J}];
 529             vdwjidx{J}B        = 2*vdwtype[jnrB+{J}];
 530             vdwjidx{J}C        = 2*vdwtype[jnrC+{J}];
 531             vdwjidx{J}D        = 2*vdwtype[jnrD+{J}];
 532             /*     #endfor */
 533             /* #endif */
 534
 535             /* #if 'Force' in KERNEL_VF and not 'Particle' in GEOMETRY_I */
 536             /*     #for J in PARTICLES_J */
 537             fjx{J}             = _mm256_setzero_pd();
 538             fjy{J}             = _mm256_setzero_pd();
 539             fjz{J}             = _mm256_setzero_pd();
 540             /*     #endfor */
 541             /* #endif */
 542
 543             /* #for I,J in PAIRS_IJ */
 544
 545             /**************************
 546              * CALCULATE INTERACTIONS *
 547              **************************/
 548
 549             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 550             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 551             /*         ## We always calculate rinv/rinvsq above to enable pipelineing in compilers (performance tested on x86) */
 552             if (gmx_mm256_any_lt(rsq{I}{J},rcutoff2))
 553             {
 554                 /*     #if 0    ## this and the next two lines is a hack to maintain auto-indentation in template file */
 555             }
 556             /*         #endif */
 557             /*         #define INNERFLOPS INNERFLOPS+1 */
 558             /*     #endif */
 559
 560             /*     #if 'r' in INTERACTION_FLAGS[I][J] */
 561             r{I}{J}              = _mm256_mul_pd(rsq{I}{J},rinv{I}{J});
 562             /*         #if ROUND == 'Epilogue' */
 563             r{I}{J}              = _mm256_andnot_pd(dummy_mask,r{I}{J});
 564             /*             #define INNERFLOPS INNERFLOPS+1 */
 565             /*         #endif */
 566             /*         #define INNERFLOPS INNERFLOPS+1 */
 567             /*     #endif */
 568
 569             /*     ## For water geometries we already loaded parameters at the start of the kernel */
 570             /*     #if not 'Water' in GEOMETRY_J */
 571             /* Compute parameters for interactions between i and j atoms */
 572             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 573             qq{I}{J}             = _mm256_mul_pd(iq{I},jq{J});
 574             /*             #define INNERFLOPS INNERFLOPS+1 */
 575             /*         #endif */
 576             /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 577             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr{I}+vdwjidx{J}A,
 578                                             vdwioffsetptr{I}+vdwjidx{J}B,
 579                                             vdwioffsetptr{I}+vdwjidx{J}C,
 580                                             vdwioffsetptr{I}+vdwjidx{J}D,
 581                                             &c6_{I}{J},&c12_{I}{J});
 582
 583             /*           #if 'LJEwald' in KERNEL_VDW */
 584             c6grid_{I}{J}       = gmx_mm256_load_4real_swizzle_pd(vdwgridioffsetptr{I}+vdwjidx{J}A,
 585                                                                   vdwgridioffsetptr{I}+vdwjidx{J}B,
 586                                                                   vdwgridioffsetptr{I}+vdwjidx{J}C,
 587                                                                   vdwgridioffsetptr{I}+vdwjidx{J}D);
 588             /*           #endif */
 589             /*         #endif */
 590             /*     #endif */
 591
 592             /*     #if 'table' in INTERACTION_FLAGS[I][J] */
 593             /* Calculate table index by multiplying r with table scale and truncate to integer */
 594             rt               = _mm256_mul_pd(r{I}{J},vftabscale);
 595             vfitab           = _mm256_cvttpd_epi32(rt);
 596             vfeps            = _mm256_sub_pd(rt,_mm256_round_pd(rt, _MM_FROUND_FLOOR));
 597             /*         #define INNERFLOPS INNERFLOPS+4                          */
 598             /*         #if 'Table' in KERNEL_ELEC and 'Table' in KERNEL_VDW     */
 599             /*             ## 3 tables, 4 bytes per point: multiply index by 12 */
 600             vfitab           = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2);
 601             /*         #elif 'Table' in KERNEL_ELEC                             */
 602             /*             ## 1 table, 4 bytes per point: multiply index by 4   */
 603             vfitab           = _mm_slli_epi32(vfitab,2);
 604             /*         #elif 'Table' in KERNEL_VDW                              */
 605             /*             ## 2 tables, 4 bytes per point: multiply index by 8  */
 606             vfitab           = _mm_slli_epi32(vfitab,3);
 607             /*         #endif                                                   */
 608             /*     #endif */
 609
 610             /*     ## ELECTROSTATIC INTERACTIONS */
 611             /*     #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 612
 613             /*         #if KERNEL_ELEC=='Coulomb' */
 614
 615             /* COULOMB ELECTROSTATICS */
 616             velec            = _mm256_mul_pd(qq{I}{J},rinv{I}{J});
 617             /*             #define INNERFLOPS INNERFLOPS+1 */
 618             /*             #if 'Force' in KERNEL_VF */
 619             felec            = _mm256_mul_pd(velec,rinvsq{I}{J});
 620             /*                 #define INNERFLOPS INNERFLOPS+1 */
 621             /*             #endif */
 622
 623             /*         #elif KERNEL_ELEC=='ReactionField' */
 624
 625             /* REACTION-FIELD ELECTROSTATICS */
 626             /*             #if 'Potential' in KERNEL_VF */
 627             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_add_pd(rinv{I}{J},_mm256_mul_pd(krf,rsq{I}{J})),crf));
 628             /*                 #define INNERFLOPS INNERFLOPS+4 */
 629             /*             #endif */
 630             /*             #if 'Force' in KERNEL_VF */
 631             felec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_mul_pd(rinv{I}{J},rinvsq{I}{J}),krf2));
 632             /*                 #define INNERFLOPS INNERFLOPS+3 */
 633             /*             #endif */
 634
 635             /*         #elif KERNEL_ELEC=='GeneralizedBorn' */
 636
 637             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 638             isaprod          = _mm256_mul_pd(isai{I},isaj{J});
 639             gbqqfactor       = _mm256_xor_pd(signbit,_mm256_mul_pd(qq{I}{J},_mm256_mul_pd(isaprod,gbinvepsdiff)));
 640             gbscale          = _mm256_mul_pd(isaprod,gbtabscale);
 641             /*             #define INNERFLOPS INNERFLOPS+5 */
 642
 643             /* Calculate generalized born table index - this is a separate table from the normal one,
 644              * but we use the same procedure by multiplying r with scale and truncating to integer.
 645              */
 646             rt               = _mm256_mul_pd(r{I}{J},gbscale);
 647             gbitab           = _mm256_cvttpd_epi32(rt);
 648             gbeps            = _mm256_sub_pd(rt,_mm256_round_pd(rt, _MM_FROUND_FLOOR));
 649             gbitab           = _mm_slli_epi32(gbitab,2);
 650             Y                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,0) );
 651             F                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,1) );
 652             G                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,2) );
 653             H                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,3) );
 654             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 655             Heps             = _mm256_mul_pd(gbeps,H);
 656             Fp               = _mm256_add_pd(F,_mm256_mul_pd(gbeps,_mm256_add_pd(G,Heps)));
 657             VV               = _mm256_add_pd(Y,_mm256_mul_pd(gbeps,Fp));
 658             vgb              = _mm256_mul_pd(gbqqfactor,VV);
 659             /*             #define INNERFLOPS INNERFLOPS+10 */
 660
 661             /*             #if 'Force' in KERNEL_VF */
 662             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 663             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
 664             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r{I}{J})));
 665             /*                 #if ROUND == 'Epilogue' */
 666             dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
 667             /*                 #endif */
 668             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
 669             /*                 #if ROUND == 'Loop' */
 670             fjptrA           = dvda+jnrA;
 671             fjptrB           = dvda+jnrB;
 672             fjptrC           = dvda+jnrC;
 673             fjptrD           = dvda+jnrD;
 674             /*                 #else */
 675             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
 676             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;
 677             fjptrB             = (jnrlistB>=0) ? dvda+jnrB : scratch;
 678             fjptrC             = (jnrlistC>=0) ? dvda+jnrC : scratch;
 679             fjptrD             = (jnrlistD>=0) ? dvda+jnrD : scratch;
 680             /*                 #endif */
 681             gmx_mm256_increment_4real_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
 682                                                  _mm256_mul_pd(dvdatmp,_mm256_mul_pd(isaj{J},isaj{J})));
 683             /*                 #define INNERFLOPS INNERFLOPS+12 */
 684             /*             #endif */
 685             velec            = _mm256_mul_pd(qq{I}{J},rinv{I}{J});
 686             /*                 #define INNERFLOPS INNERFLOPS+1 */
 687             /*             #if 'Force' in KERNEL_VF */
 688             felec            = _mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(velec,rinv{I}{J}),fgb),rinv{I}{J});
 689             /*                 #define INNERFLOPS INNERFLOPS+3 */
 690             /*             #endif */
 691
 692             /*         #elif KERNEL_ELEC=='Ewald' */
 693             /* EWALD ELECTROSTATICS */
 694
 695             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 696             ewrt             = _mm256_mul_pd(r{I}{J},ewtabscale);
 697             ewitab           = _mm256_cvttpd_epi32(ewrt);
 698             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 699             /*             #define INNERFLOPS INNERFLOPS+4 */
 700             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_ELEC=='PotentialSwitch' */
 701             ewitab           = _mm_slli_epi32(ewitab,2);
 702             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 703             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 704             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 705             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 706             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 707             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 708             /*                 #define INNERFLOPS INNERFLOPS+2 */
 709             /*                 #if KERNEL_MOD_ELEC=='PotentialShift' */
 710             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 711             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_sub_pd(rinv{I}{J},sh_ewald),velec));
 712             /*                     #define INNERFLOPS INNERFLOPS+7 */
 713             /*                 #else */
 714             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 715             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(rinv{I}{J},velec));
 716             /*                     #define INNERFLOPS INNERFLOPS+6 */
 717             /*                 #endif */
 718             /*                 #if 'Force' in KERNEL_VF */
 719             felec            = _mm256_mul_pd(_mm256_mul_pd(qq{I}{J},rinv{I}{J}),_mm256_sub_pd(rinvsq{I}{J},felec));
 720             /*                      #define INNERFLOPS INNERFLOPS+3 */
 721             /*                 #endif */
 722             /*             #elif KERNEL_VF=='Force' */
 723             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
 724                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
 725                                             &ewtabF,&ewtabFn);
 726             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
 727             felec            = _mm256_mul_pd(_mm256_mul_pd(qq{I}{J},rinv{I}{J}),_mm256_sub_pd(rinvsq{I}{J},felec));
 728             /*                 #define INNERFLOPS INNERFLOPS+7 */
 729             /*             #endif */
 730
 731             /*         #elif KERNEL_ELEC=='CubicSplineTable' */
 732
 733             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 734             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 735             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 736             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 737             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 738             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 739             Heps             = _mm256_mul_pd(vfeps,H);
 740             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 741             /*             #define INNERFLOPS INNERFLOPS+4 */
 742             /*             #if 'Potential' in KERNEL_VF */
 743             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 744             velec            = _mm256_mul_pd(qq{I}{J},VV);
 745             /*                 #define INNERFLOPS INNERFLOPS+3 */
 746             /*             #endif */
 747             /*             #if 'Force' in KERNEL_VF */
 748             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 749             felec            = _mm256_xor_pd(signbit,_mm256_mul_pd(_mm256_mul_pd(qq{I}{J},FF),_mm256_mul_pd(vftabscale,rinv{I}{J})));
 750             /*                 #define INNERFLOPS INNERFLOPS+7 */
 751             /*             #endif */
 752             /*         #endif */
 753             /*         ## End of check for electrostatics interaction forms */
 754             /*     #endif */
 755             /*     ## END OF ELECTROSTATIC INTERACTION CHECK FOR PAIR I-J */
 756
 757             /*     #if 'vdw' in INTERACTION_FLAGS[I][J] */
 758
 759             /*         #if KERNEL_VDW=='LennardJones' */
 760
 761             /* LENNARD-JONES DISPERSION/REPULSION */
 762
 763             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq{I}{J},rinvsq{I}{J}),rinvsq{I}{J});
 764             /*             #define INNERFLOPS INNERFLOPS+2 */
 765             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_VDW=='PotentialSwitch' */
 766             vvdw6            = _mm256_mul_pd(c6_{I}{J},rinvsix);
 767             vvdw12           = _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(rinvsix,rinvsix));
 768             /*                 #define INNERFLOPS INNERFLOPS+3 */
 769             /*                 #if KERNEL_MOD_VDW=='PotentialShift' */
 770             vvdw             = _mm256_sub_pd(_mm256_mul_pd( _mm256_sub_pd(vvdw12 , _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
 771                                           _mm256_mul_pd( _mm256_sub_pd(vvdw6,_mm256_mul_pd(c6_{I}{J},sh_vdw_invrcut6)),one_sixth));
 772             /*                     #define INNERFLOPS INNERFLOPS+8 */
 773             /*                 #else */
 774             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
 775             /*                     #define INNERFLOPS INNERFLOPS+3 */
 776             /*                 #endif */
 777             /*                 ## Check for force inside potential check, i.e. this means we already did the potential part */
 778             /*                 #if 'Force' in KERNEL_VF */
 779             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq{I}{J});
 780             /*                     #define INNERFLOPS INNERFLOPS+2 */
 781             /*                 #endif */
 782             /*             #elif KERNEL_VF=='Force' */
 783             /*                 ## Force-only LennardJones makes it possible to save 1 flop (they do add up...) */
 784             fvdw             = _mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(c12_{I}{J},rinvsix),c6_{I}{J}),_mm256_mul_pd(rinvsix,rinvsq{I}{J}));
 785             /*                 #define INNERFLOPS INNERFLOPS+4 */
 786             /*             #endif */
 787
 788             /*         #elif KERNEL_VDW=='CubicSplineTable' */
 789
 790             /* CUBIC SPLINE TABLE DISPERSION */
 791             /*             #if 'Table' in KERNEL_ELEC */
 792             vfitab           = _mm_add_epi32(vfitab,ifour);
 793             /*             #endif                     */
 794             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 795             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 796             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 797             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 798             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 799             Heps             = _mm256_mul_pd(vfeps,H);
 800             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 801             /*             #define INNERFLOPS INNERFLOPS+4 */
 802             /*             #if 'Potential' in KERNEL_VF */
 803             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 804             vvdw6            = _mm256_mul_pd(c6_{I}{J},VV);
 805             /*                 #define INNERFLOPS INNERFLOPS+3 */
 806             /*             #endif */
 807             /*             #if 'Force' in KERNEL_VF */
 808             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 809             fvdw6            = _mm256_mul_pd(c6_{I}{J},FF);
 810             /*                 #define INNERFLOPS INNERFLOPS+4 */
 811             /*             #endif */
 812
 813             /* CUBIC SPLINE TABLE REPULSION */
 814             vfitab           = _mm_add_epi32(vfitab,ifour);
 815             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 816             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 817             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 818             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 819             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 820             Heps             = _mm256_mul_pd(vfeps,H);
 821             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 822             /*             #define INNERFLOPS INNERFLOPS+4 */
 823             /*             #if 'Potential' in KERNEL_VF */
 824             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 825             vvdw12           = _mm256_mul_pd(c12_{I}{J},VV);
 826             /*                 #define INNERFLOPS INNERFLOPS+3 */
 827             /*             #endif */
 828             /*             #if 'Force' in KERNEL_VF */
 829             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 830             fvdw12           = _mm256_mul_pd(c12_{I}{J},FF);
 831             /*                 #define INNERFLOPS INNERFLOPS+5 */
 832             /*             #endif */
 833             /*             #if 'Potential' in KERNEL_VF */
 834             vvdw             = _mm256_add_pd(vvdw12,vvdw6);
 835             /*                 #define INNERFLOPS INNERFLOPS+1 */
 836             /*             #endif */
 837             /*             #if 'Force' in KERNEL_VF */
 838             fvdw             = _mm256_xor_pd(signbit,_mm256_mul_pd(_mm256_add_pd(fvdw6,fvdw12),_mm256_mul_pd(vftabscale,rinv{I}{J})));
 839             /*                 #define INNERFLOPS INNERFLOPS+4 */
 840             /*             #endif */
 841
 842             /*         #elif KERNEL_VDW=='LJEwald' */
 843
 844             /* Analytical LJ-PME */
 845             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq{I}{J},rinvsq{I}{J}),rinvsq{I}{J});
 846             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq{I}{J});
 847             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
 848             exponent         = avx256_exp_d(ewcljrsq);
 849             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 850             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
 851             /*                 #define INNERFLOPS INNERFLOPS+11 */
 852             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_VDW=='PotentialSwitch' */
 853             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 854             vvdw6            = _mm256_mul_pd(_mm256_sub_pd(c6_{I}{J},_mm256_mul_pd(c6grid_{I}{J},_mm256_sub_pd(one,poly))),rinvsix);
 855             vvdw12           = _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(rinvsix,rinvsix));
 856             /*                     #define INNERFLOPS INNERFLOPS+6 */
 857             /*                 #if KERNEL_MOD_VDW=='PotentialShift' */
 858             vvdw             = _mm256_sub_pd(_mm256_mul_pd( _mm256_sub_pd(vvdw12 , _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(sh_vdw_invrcut6,sh_vdw_invrcut6))), one_twelfth) ,
 859                                           _mm256_mul_pd( _mm256_sub_pd(vvdw6,_mm256_add_pd(_mm256_mul_pd(c6_{I}{J},sh_vdw_invrcut6),_mm256_mul_pd(c6grid_{I}{J},sh_lj_ewald))),one_sixth));
 860             /*                     #define INNERFLOPS INNERFLOPS+10 */
 861             /*                 #else */
 862             vvdw             = _mm256_sub_pd(_mm256_mul_pd(vvdw12,one_twelfth),_mm256_mul_pd(vvdw6,one_sixth));
 863             /*                 #define INNERFLOPS INNERFLOPS+6 */
 864             /*                 #endif */
 865             /*                  ## Check for force inside potential check, i.e. this means we already did the potential part */
 866             /*                  #if 'Force' in KERNEL_VF */
 867             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 868             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,_mm256_sub_pd(vvdw6,_mm256_mul_pd(_mm256_mul_pd(c6grid_{I}{J},one_sixth),_mm256_mul_pd(exponent,ewclj6)))),rinvsq{I}{J});
 869             /*                 #define INNERFLOPS INNERFLOPS+6 */
 870             /*                  #endif */
 871             /*              #elif KERNEL_VF=='Force' */
 872             /* f6A = 6 * C6grid * (1 - poly) */
 873             f6A              = _mm256_mul_pd(c6grid_{I}{J},_mm256_sub_pd(one,poly));
 874             /* f6B = C6grid * exponent * beta^6 */
 875             f6B              = _mm256_mul_pd(_mm256_mul_pd(c6grid_{I}{J},one_sixth),_mm256_mul_pd(exponent,ewclj6));
 876             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 877             fvdw              = _mm256_mul_pd(_mm256_add_pd(_mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(c12_{I}{J},rinvsix),_mm256_sub_pd(c6_{I}{J},f6A)),rinvsix),f6B),rinvsq{I}{J});
 878             /*                 #define INNERFLOPS INNERFLOPS+11 */
 879             /*              #endif */
 880             /*         #endif */
 881             /*         ## End of check for vdw interaction forms */
 882             /*     #endif */
 883             /*     ## END OF VDW INTERACTION CHECK FOR PAIR I-J */
 884
 885             /*     #if 'switch' in INTERACTION_FLAGS[I][J] */
 886             d                = _mm256_sub_pd(r{I}{J},rswitch);
 887             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 888             d2               = _mm256_mul_pd(d,d);
 889             sw               = _mm256_add_pd(one,_mm256_mul_pd(d2,_mm256_mul_pd(d,_mm256_add_pd(swV3,_mm256_mul_pd(d,_mm256_add_pd(swV4,_mm256_mul_pd(d,swV5)))))));
 890             /*         #define INNERFLOPS INNERFLOPS+10 */
 891
 892             /*         #if 'Force' in KERNEL_VF */
 893             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 894             /*             #define INNERFLOPS INNERFLOPS+5 */
 895             /*         #endif */
 896
 897             /* Evaluate switch function */
 898             /*         #if 'Force' in KERNEL_VF */
 899             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 900             /*             #if 'electrostatics' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_ELEC=='PotentialSwitch' */
 901             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv{I}{J},_mm256_mul_pd(velec,dsw)) );
 902             /*                 #define INNERFLOPS INNERFLOPS+4 */
 903             /*             #endif */
 904             /*             #if 'vdw' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_VDW=='PotentialSwitch' */
 905             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv{I}{J},_mm256_mul_pd(vvdw,dsw)) );
 906             /*                 #define INNERFLOPS INNERFLOPS+4 */
 907             /*             #endif */
 908             /*         #endif */
 909             /*         #if 'Potential' in KERNEL_VF */
 910             /*             #if 'electrostatics' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_ELEC=='PotentialSwitch' */
 911             velec            = _mm256_mul_pd(velec,sw);
 912             /*                 #define INNERFLOPS INNERFLOPS+1 */
 913             /*             #endif */
 914             /*             #if 'vdw' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_VDW=='PotentialSwitch' */
 915             vvdw             = _mm256_mul_pd(vvdw,sw);
 916             /*                 #define INNERFLOPS INNERFLOPS+1 */
 917             /*             #endif */
 918             /*         #endif */
 919             /*     #endif */
 920             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 921             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 922             cutoff_mask      = _mm256_cmp_pd(rsq{I}{J},rcutoff2,_CMP_LT_OQ);
 923             /*         #define INNERFLOPS INNERFLOPS+1 */
 924             /*     #endif */
 925
 926             /*     #if 'Potential' in KERNEL_VF */
 927             /* Update potential sum for this i atom from the interaction with this j atom. */
 928             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 929             /*             #if 'exactcutoff' in INTERACTION_FLAGS[I][J] */
 930             velec            = _mm256_and_pd(velec,cutoff_mask);
 931             /*                 #define INNERFLOPS INNERFLOPS+1 */
 932             /*             #endif                                       */
 933             /*             #if ROUND == 'Epilogue' */
 934             velec            = _mm256_andnot_pd(dummy_mask,velec);
 935             /*             #endif */
 936             velecsum         = _mm256_add_pd(velecsum,velec);
 937             /*             #define INNERFLOPS INNERFLOPS+1 */
 938             /*             #if KERNEL_ELEC=='GeneralizedBorn' */
 939             /*             #if 'exactcutoff' in INTERACTION_FLAGS[I][J] */
 940             vgb              = _mm256_and_pd(vgb,cutoff_mask);
 941             /*                 #define INNERFLOPS INNERFLOPS+1 */
 942             /*             #endif                                       */
 943             /*             #if ROUND == 'Epilogue' */
 944             vgb              = _mm256_andnot_pd(dummy_mask,vgb);
 945             /*             #endif */
 946             vgbsum           = _mm256_add_pd(vgbsum,vgb);
 947             /*                 #define INNERFLOPS INNERFLOPS+1 */
 948             /*             #endif */
 949             /*         #endif */
 950             /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 951             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 952             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 953             vvdw             = _mm256_and_pd(vvdw,cutoff_mask);
 954             /*                 #define INNERFLOPS INNERFLOPS+1 */
 955             /*             #endif                                       */
 956             /*             #if ROUND == 'Epilogue' */
 957             vvdw             = _mm256_andnot_pd(dummy_mask,vvdw);
 958             /*             #endif */
 959             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 960             /*             #define INNERFLOPS INNERFLOPS+1 */
 961             /*         #endif */
 962             /*     #endif */
 963
 964             /*     #if 'Force' in KERNEL_VF */
 965
 966             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] and 'vdw' in INTERACTION_FLAGS[I][J] */
 967             fscal            = _mm256_add_pd(felec,fvdw);
 968             /*             #define INNERFLOPS INNERFLOPS+1 */
 969             /*         #elif 'electrostatics' in INTERACTION_FLAGS[I][J] */
 970             fscal            = felec;
 971             /*         #elif 'vdw' in INTERACTION_FLAGS[I][J] */
 972             fscal            = fvdw;
 973             /*        #endif */
 974
 975             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 976             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 977             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 978             /*                 #define INNERFLOPS INNERFLOPS+1 */
 979             /*             #endif                                       */
 980
 981             /*             #if ROUND == 'Epilogue' */
 982             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 983             /*             #endif */
 984
 985             /* Calculate temporary vectorial force */
 986             tx               = _mm256_mul_pd(fscal,dx{I}{J});
 987             ty               = _mm256_mul_pd(fscal,dy{I}{J});
 988             tz               = _mm256_mul_pd(fscal,dz{I}{J});
 989
 990             /* Update vectorial force */
 991             fix{I}             = _mm256_add_pd(fix{I},tx);
 992             fiy{I}             = _mm256_add_pd(fiy{I},ty);
 993             fiz{I}             = _mm256_add_pd(fiz{I},tz);
 994             /*             #define INNERFLOPS INNERFLOPS+6 */
 995
 996             /* #if GEOMETRY_I == 'Particle'             */
 997             /*     #if ROUND == 'Loop' */
 998             fjptrA             = f+j_coord_offsetA;
 999             fjptrB             = f+j_coord_offsetB;
1000             fjptrC             = f+j_coord_offsetC;
1001             fjptrD             = f+j_coord_offsetD;
1002             /*     #else */
1003             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1004             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1005             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1006             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1007             /*     #endif */
1008             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,tx,ty,tz);
1009             /*     #define INNERFLOPS INNERFLOPS+3      */
1010             /* #else                                    */
1011             fjx{J}             = _mm256_add_pd(fjx{J},tx);
1012             fjy{J}             = _mm256_add_pd(fjy{J},ty);
1013             fjz{J}             = _mm256_add_pd(fjz{J},tz);
1014             /*     #define INNERFLOPS INNERFLOPS+3      */
1015             /* #endif                                   */
1016
1017             /*     #endif */
1018
1019             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
1020             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
1021             /*         #if 0    ## This and next two lines is a hack to maintain indentation in template file */
1022             {
1023                 /*     #endif */
1024             }
1025             /*     #endif */
1026             /*    ## End of check for the interaction being outside the cutoff */
1027
1028             /* #endfor */
1029             /* ## End of loop over i-j interaction pairs */
1030
1031             /* #if GEOMETRY_I != 'Particle' */
1032             /*     #if ROUND == 'Loop' */
1033             fjptrA             = f+j_coord_offsetA;
1034             fjptrB             = f+j_coord_offsetB;
1035             fjptrC             = f+j_coord_offsetC;
1036             fjptrD             = f+j_coord_offsetD;
1037             /*     #else */
1038             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1039             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1040             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1041             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1042             /*     #endif */
1043             /* #endif */
1044
1045             /* #if 'Water' in GEOMETRY_I and GEOMETRY_J == 'Particle' */
1046             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1047             /*     #define INNERFLOPS INNERFLOPS+3      */
1048             /* #elif GEOMETRY_J == 'Water3'             */
1049             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1050                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1051             /*     #define INNERFLOPS INNERFLOPS+9      */
1052             /* #elif GEOMETRY_J == 'Water4'             */
1053             /*     #if 0 in PARTICLES_J                 */
1054             gmx_mm256_decrement_4rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1055                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
1056                                                       fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1057             /*     #define INNERFLOPS INNERFLOPS+12     */
1058             /*     #else                                */
1059             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
1060                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1061             /*     #define INNERFLOPS INNERFLOPS+9      */
1062             /*     #endif                               */
1063             /* #endif                                   */
1064
1065             /* Inner loop uses {INNERFLOPS} flops */
1066         }
1067
1068         /* #endfor */
1069
1070         /* End of innermost loop */
1071
1072         /* #if 'Force' in KERNEL_VF */
1073         /*     #if GEOMETRY_I == 'Particle'            */
1074         gmx_mm256_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
1075                                                  f+i_coord_offset,fshift+i_shift_offset);
1076         /*         #define OUTERFLOPS OUTERFLOPS+6     */
1077         /*     #elif GEOMETRY_I == 'Water3'            */
1078         gmx_mm256_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1079                                                  f+i_coord_offset,fshift+i_shift_offset);
1080         /*         #define OUTERFLOPS OUTERFLOPS+18    */
1081         /*     #elif GEOMETRY_I == 'Water4'            */
1082         /*         #if 0 in PARTICLES_I                */
1083         gmx_mm256_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1084                                                  f+i_coord_offset,fshift+i_shift_offset);
1085         /*             #define OUTERFLOPS OUTERFLOPS+24    */
1086         /*         #else                               */
1087         gmx_mm256_update_iforce_3atom_swizzle_pd(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1088                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
1089         /*             #define OUTERFLOPS OUTERFLOPS+18    */
1090         /*         #endif                              */
1091         /*     #endif                                  */
1092         /* #endif                                      */
1093
1094         /* #if 'Potential' in KERNEL_VF */
1095         ggid                        = gid[iidx];
1096         /* Update potential energies */
1097         /*     #if KERNEL_ELEC != 'None' */
1098         gmx_mm256_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
1099         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1100         /*     #endif */
1101         /*     #if 'GeneralizedBorn' in KERNEL_ELEC */
1102         gmx_mm256_update_1pot_pd(vgbsum,kernel_data->energygrp_polarization+ggid);
1103         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1104         /*     #endif */
1105         /*     #if KERNEL_VDW != 'None' */
1106         gmx_mm256_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
1107         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1108         /*     #endif */
1109         /* #endif */
1110         /*     #if 'GeneralizedBorn' in KERNEL_ELEC and 'Force' in KERNEL_VF */
1111         dvdasum = _mm256_mul_pd(dvdasum, _mm256_mul_pd(isai{I},isai{I}));
1112         gmx_mm256_update_1pot_pd(dvdasum,dvda+inr);
1113         /*     #endif */
1114
1115         /* Increment number of inner iterations */
1116         inneriter                  += j_index_end - j_index_start;
1117
1118         /* Outer loop uses {OUTERFLOPS} flops */
1119     }
1120
1121     /* Increment number of outer iterations */
1122     outeriter        += nri;
1123
1124     /* Update outer/inner flops */
1125     /* ## NB: This is not important, it just affects the flopcount. However, since our preprocessor is */
1126     /* ## primitive and replaces aggressively even in strings inside these directives, we need to      */
1127     /* ## assemble the main part of the name (containing KERNEL/ELEC/VDW) directly in the source.      */
1128     /* #if GEOMETRY_I == 'Water3'            */
1129     /*     #define ISUFFIX '_W3'             */
1130     /* #elif GEOMETRY_I == 'Water4'          */
1131     /*     #define ISUFFIX '_W4'             */
1132     /* #else                                 */
1133     /*     #define ISUFFIX ''                */
1134     /* #endif                                */
1135     /* #if GEOMETRY_J == 'Water3'            */
1136     /*     #define JSUFFIX 'W3'              */
1137     /* #elif GEOMETRY_J == 'Water4'          */
1138     /*     #define JSUFFIX 'W4'              */
1139     /* #else                                 */
1140     /*     #define JSUFFIX ''                */
1141     /* #endif                                */
1142     /* #if 'PotentialAndForce' in KERNEL_VF  */
1143     /*     #define VFSUFFIX  '_VF'           */
1144     /* #elif 'Potential' in KERNEL_VF        */
1145     /*     #define VFSUFFIX '_V'             */
1146     /* #else                                 */
1147     /*     #define VFSUFFIX '_F'             */
1148     /* #endif                                */
1149
1150     /* #if KERNEL_ELEC != 'None' and KERNEL_VDW != 'None' */
1151     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1152     /* #elif KERNEL_ELEC != 'None' */
1153     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1154     /* #else */
1155     inc_nrnb(nrnb,eNR_NBKERNEL_VDW{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1156     /* #endif  */
1157 }