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, 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 #ifdef HAVE_CONFIG_H
  40 #include <config.h>
  41 #endif
  42
  43 #include <math.h>
  44
  45 #include "../nb_kernel.h"
  46 #include "types/simple.h"
  47 #include "gromacs/math/vec.h"
  48 #include "nrnb.h"
  49
  50 #include "gromacs/simd/math_x86_avx_256_double.h"
  51 #include "kernelutil_x86_avx_256_double.h"
  52 /* #endif */
  53
  54 /* ## List of variables set by the generating script:                                    */
  55 /* ##                                                                                    */
  56 /* ## Setttings that apply to the entire kernel:                                         */
  57 /* ## KERNEL_ELEC:           String, choice for electrostatic interactions               */
  58 /* ## KERNEL_VDW:            String, choice for van der Waals interactions               */
  59 /* ## KERNEL_NAME:           String, name of this kernel                                 */
  60 /* ## KERNEL_VF:             String telling if we calculate potential, force, or both    */
  61 /* ## GEOMETRY_I/GEOMETRY_J: String, name of each geometry, e.g. 'Water3' or '1Particle' */
  62 /* ##                                                                                    */
  63 /* ## Setttings that apply to particles in the outer (I) or inner (J) loops:             */
  64 /* ## PARTICLES_I[]/         Arrays with lists of i/j particles to use in kernel. It is  */
  65 /* ## PARTICLES_J[]:         just [0] for particle geometry, but can be longer for water */
  66 /* ## PARTICLES_ELEC_I[]/    Arrays with lists of i/j particle that have electrostatics  */
  67 /* ## PARTICLES_ELEC_J[]:    interactions that should be calculated in this kernel.      */
  68 /* ## PARTICLES_VDW_I[]/     Arrays with the list of i/j particle that have VdW          */
  69 /* ## PARTICLES_VDW_J[]:     interactions that should be calculated in this kernel.      */
  70 /* ##                                                                                    */
  71 /* ## Setttings for pairs of interactions (e.g. 2nd i particle against 1st j particle)   */
  72 /* ## PAIRS_IJ[]:            Array with (i,j) tuples of pairs for which interactions     */
  73 /* ##                        should be calculated in this kernel. Zero-charge particles  */
  74 /* ##                        do not have interactions with particles without vdw, and    */
  75 /* ##                        Vdw-only interactions are not evaluated in a no-vdw-kernel. */
  76 /* ## INTERACTION_FLAGS[][]: 2D matrix, dimension e.g. 3*3 for water-water interactions. */
  77 /* ##                        For each i-j pair, the element [I][J] is a list of strings  */
  78 /* ##                        defining properties/flags of this interaction. Examples     */
  79 /* ##                        include 'electrostatics'/'vdw' if that type of interaction  */
  80 /* ##                        should be evaluated, 'rsq'/'rinv'/'rinvsq' if those values  */
  81 /* ##                        are needed, and 'exactcutoff' or 'shift','switch' to        */
  82 /* ##                        decide if the force/potential should be modified. This way  */
  83 /* ##                        we only calculate values absolutely needed for each case.   */
  84
  85 /* ## Calculate the size and offset for (merged/interleaved) table data */
  86
  87 /*
  88  * Gromacs nonbonded kernel:   {KERNEL_NAME}
  89  * Electrostatics interaction: {KERNEL_ELEC}
  90  * VdW interaction:            {KERNEL_VDW}
  91  * Geometry:                   {GEOMETRY_I}-{GEOMETRY_J}
  92  * Calculate force/pot:        {KERNEL_VF}
  93  */
  94 void
  95 {KERNEL_NAME}
  96                     (t_nblist                    * gmx_restrict       nlist,
  97                      rvec                        * gmx_restrict          xx,
  98                      rvec                        * gmx_restrict          ff,
  99                      t_forcerec                  * gmx_restrict          fr,
 100                      t_mdatoms                   * gmx_restrict     mdatoms,
 101                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 102                      t_nrnb                      * gmx_restrict        nrnb)
 103 {
 104     /* ## Not all variables are used for all kernels, but any optimizing compiler fixes that, */
 105     /* ## so there is no point in going to extremes to exclude variables that are not needed. */
 106     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 107      * just 0 for non-waters.
 108      * Suffixes A,B,C,D refer to j loop unrolling done with AVX, e.g. for the four different
 109      * jnr indices corresponding to data put in the four positions in the SIMD register.
 110      */
 111     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 112     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 113     int              jnrA,jnrB,jnrC,jnrD;
 114     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 115     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 116     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 117     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 118     real             rcutoff_scalar;
 119     real             *shiftvec,*fshift,*x,*f;
 120     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 121     real             scratch[4*DIM];
 122     __m256d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 123     /* #for I in PARTICLES_I */
 124     real *           vdwioffsetptr{I};
 125     /* #if 'LJEwald' in KERNEL_VDW */
 126     real *           vdwgridioffsetptr{I};
 127     /* #endif                      */
 128     __m256d          ix{I},iy{I},iz{I},fix{I},fiy{I},fiz{I},iq{I},isai{I};
 129     /* #endfor */
 130     /* #for J in PARTICLES_J */
 131     int              vdwjidx{J}A,vdwjidx{J}B,vdwjidx{J}C,vdwjidx{J}D;
 132     __m256d          jx{J},jy{J},jz{J},fjx{J},fjy{J},fjz{J},jq{J},isaj{J};
 133     /* #endfor */
 134     /* #for I,J in PAIRS_IJ */
 135     __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};
 136     /* #endfor */
 137     /* #if KERNEL_ELEC != 'None' */
 138     __m256d          velec,felec,velecsum,facel,crf,krf,krf2;
 139     real             *charge;
 140     /* #endif */
 141     /* #if 'GeneralizedBorn' in KERNEL_ELEC */
 142     __m128i          gbitab;
 143     __m256d          vgb,fgb,vgbsum,dvdasum,gbscale,gbtabscale,isaprod,gbqqfactor,gbinvepsdiff,gbeps,dvdatmp;
 144     __m256d          minushalf = _mm256_set1_pd(-0.5);
 145     real             *invsqrta,*dvda,*gbtab;
 146     /* #endif */
 147     /* #if KERNEL_VDW != 'None' */
 148     int              nvdwtype;
 149     __m256d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 150     int              *vdwtype;
 151     real             *vdwparam;
 152     __m256d          one_sixth   = _mm256_set1_pd(1.0/6.0);
 153     __m256d          one_twelfth = _mm256_set1_pd(1.0/12.0);
 154     /* #endif */
 155     /* #if 'Table' in KERNEL_ELEC or 'GeneralizedBorn' in KERNEL_ELEC or 'Table' in KERNEL_VDW */
 156     __m128i          vfitab;
 157     __m128i          ifour       = _mm_set1_epi32(4);
 158     __m256d          rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
 159     real             *vftab;
 160     /* #endif */
 161     /* #if 'LJEwald' in KERNEL_VDW */
 162     /* #for I,J in PAIRS_IJ */
 163     __m256d           c6grid_{I}{J};
 164     /* #endfor */
 165     real             *vdwgridparam;
 166     __m256d           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 167     __m256d           one_half  = _mm256_set1_pd(0.5);
 168     __m256d           minus_one = _mm256_set1_pd(-1.0);
 169     /* #endif */
 170     /* #if 'Ewald' in KERNEL_ELEC */
 171     __m128i          ewitab;
 172     __m256d          ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 173     __m256d          beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 174     real             *ewtab;
 175     /* #endif */
 176     /* #if 'PotentialSwitch' in [KERNEL_MOD_ELEC,KERNEL_MOD_VDW] */
 177     __m256d          rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 178     real             rswitch_scalar,d_scalar;
 179     /* #endif */
 180     __m256d          dummy_mask,cutoff_mask;
 181     __m128           tmpmask0,tmpmask1;
 182     __m256d          signbit = _mm256_castsi256_pd( _mm256_set1_epi32(0x80000000) );
 183     __m256d          one     = _mm256_set1_pd(1.0);
 184     __m256d          two     = _mm256_set1_pd(2.0);
 185     x                = xx[0];
 186     f                = ff[0];
 187
 188     nri              = nlist->nri;
 189     iinr             = nlist->iinr;
 190     jindex           = nlist->jindex;
 191     jjnr             = nlist->jjnr;
 192     shiftidx         = nlist->shift;
 193     gid              = nlist->gid;
 194     shiftvec         = fr->shift_vec[0];
 195     fshift           = fr->fshift[0];
 196     /* #if KERNEL_ELEC != 'None' */
 197     facel            = _mm256_set1_pd(fr->epsfac);
 198     charge           = mdatoms->chargeA;
 199     /*     #if 'ReactionField' in KERNEL_ELEC */
 200     krf              = _mm256_set1_pd(fr->ic->k_rf);
 201     krf2             = _mm256_set1_pd(fr->ic->k_rf*2.0);
 202     crf              = _mm256_set1_pd(fr->ic->c_rf);
 203     /*     #endif */
 204     /* #endif */
 205     /* #if KERNEL_VDW != 'None' */
 206     nvdwtype         = fr->ntype;
 207     vdwparam         = fr->nbfp;
 208     vdwtype          = mdatoms->typeA;
 209     /* #endif */
 210     /* #if 'LJEwald' in KERNEL_VDW */
 211     vdwgridparam     = fr->ljpme_c6grid;
 212     sh_lj_ewald      = _mm256_set1_pd(fr->ic->sh_lj_ewald);
 213     ewclj            = _mm256_set1_pd(fr->ewaldcoeff_lj);
 214     ewclj2           = _mm256_mul_pd(minus_one,_mm256_mul_pd(ewclj,ewclj));
 215     /* #endif */
 216
 217     /* #if 'Table' in KERNEL_ELEC and 'Table' in KERNEL_VDW */
 218     vftab            = kernel_data->table_elec_vdw->data;
 219     vftabscale       = _mm256_set1_pd(kernel_data->table_elec_vdw->scale);
 220     /* #elif 'Table' in KERNEL_ELEC */
 221     vftab            = kernel_data->table_elec->data;
 222     vftabscale       = _mm256_set1_pd(kernel_data->table_elec->scale);
 223     /* #elif 'Table' in KERNEL_VDW */
 224     vftab            = kernel_data->table_vdw->data;
 225     vftabscale       = _mm256_set1_pd(kernel_data->table_vdw->scale);
 226     /* #endif */
 227
 228     /* #if 'Ewald' in KERNEL_ELEC */
 229     sh_ewald         = _mm256_set1_pd(fr->ic->sh_ewald);
 230     beta             = _mm256_set1_pd(fr->ic->ewaldcoeff_q);
 231     beta2            = _mm256_mul_pd(beta,beta);
 232     beta3            = _mm256_mul_pd(beta,beta2);
 233
 234     /*     #if KERNEL_VF=='Force' and KERNEL_MOD_ELEC!='PotentialSwitch' */
 235     ewtab            = fr->ic->tabq_coul_F;
 236     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 237     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 238     /*     #else */
 239     ewtab            = fr->ic->tabq_coul_FDV0;
 240     ewtabscale       = _mm256_set1_pd(fr->ic->tabq_scale);
 241     ewtabhalfspace   = _mm256_set1_pd(0.5/fr->ic->tabq_scale);
 242      /*     #endif */
 243     /* #endif */
 244
 245     /* #if KERNEL_ELEC=='GeneralizedBorn' */
 246     invsqrta         = fr->invsqrta;
 247     dvda             = fr->dvda;
 248     gbtabscale       = _mm256_set1_pd(fr->gbtab.scale);
 249     gbtab            = fr->gbtab.data;
 250     gbinvepsdiff     = _mm256_set1_pd((1.0/fr->epsilon_r) - (1.0/fr->gb_epsilon_solvent));
 251     /* #endif */
 252
 253     /* #if 'Water' in GEOMETRY_I */
 254     /* Setup water-specific parameters */
 255     inr              = nlist->iinr[0];
 256     /*     #for I in PARTICLES_ELEC_I */
 257     iq{I}              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+{I}]));
 258     /*     #endfor */
 259     /*     #for I in PARTICLES_VDW_I */
 260     vdwioffsetptr{I}   = vdwparam+2*nvdwtype*vdwtype[inr+{I}];
 261     /*         #if 'LJEwald' in KERNEL_VDW */
 262     vdwgridioffsetptr{I} = vdwgridparam+2*nvdwtype*vdwtype[inr+{I}];
 263     /*         #endif */
 264     /*     #endfor */
 265     /* #endif */
 266
 267     /* #if 'Water' in GEOMETRY_J */
 268     /*     #for J in PARTICLES_ELEC_J */
 269     jq{J}              = _mm256_set1_pd(charge[inr+{J}]);
 270     /*     #endfor */
 271     /*     #for J in PARTICLES_VDW_J */
 272     vdwjidx{J}A        = 2*vdwtype[inr+{J}];
 273     /*     #endfor */
 274     /*     #for I,J in PAIRS_IJ */
 275     /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 276     qq{I}{J}             = _mm256_mul_pd(iq{I},jq{J});
 277     /*         #endif */
 278     /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 279     /*             #if 'LJEwald' in KERNEL_VDW */
 280     c6_{I}{J}            = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A]);
 281     c12_{I}{J}           = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A+1]);
 282     c6grid_{I}{J}        = _mm256_set1_pd(vdwgridioffsetptr{I}[vdwjidx{J}A]);
 283     /*             #else */
 284     c6_{I}{J}            = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A]);
 285     c12_{I}{J}           = _mm256_set1_pd(vdwioffsetptr{I}[vdwjidx{J}A+1]);
 286     /*             #endif */
 287     /*         #endif */
 288     /*     #endfor */
 289     /* #endif */
 290
 291     /* #if KERNEL_MOD_ELEC!='None' or KERNEL_MOD_VDW!='None' */
 292     /*     #if KERNEL_ELEC!='None' */
 293     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 294     rcutoff_scalar   = fr->rcoulomb;
 295     /*     #else */
 296     rcutoff_scalar   = fr->rvdw;
 297     /*     #endif */
 298     rcutoff          = _mm256_set1_pd(rcutoff_scalar);
 299     rcutoff2         = _mm256_mul_pd(rcutoff,rcutoff);
 300     /* #endif */
 301
 302     /* #if KERNEL_MOD_VDW=='PotentialShift' */
 303     sh_vdw_invrcut6  = _mm256_set1_pd(fr->ic->sh_invrc6);
 304     rvdw             = _mm256_set1_pd(fr->rvdw);
 305     /* #endif */
 306
 307     /* #if 'PotentialSwitch' in [KERNEL_MOD_ELEC,KERNEL_MOD_VDW] */
 308     /*     #if KERNEL_MOD_ELEC=='PotentialSwitch'  */
 309     rswitch_scalar   = fr->rcoulomb_switch;
 310     rswitch          = _mm256_set1_pd(rswitch_scalar);
 311     /*     #else */
 312     rswitch_scalar   = fr->rvdw_switch;
 313     rswitch          = _mm256_set1_pd(rswitch_scalar);
 314     /*     #endif */
 315     /* Setup switch parameters */
 316     d_scalar         = rcutoff_scalar-rswitch_scalar;
 317     d                = _mm256_set1_pd(d_scalar);
 318     swV3             = _mm256_set1_pd(-10.0/(d_scalar*d_scalar*d_scalar));
 319     swV4             = _mm256_set1_pd( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 320     swV5             = _mm256_set1_pd( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 321     /*     #if 'Force' in KERNEL_VF */
 322     swF2             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar));
 323     swF3             = _mm256_set1_pd( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 324     swF4             = _mm256_set1_pd(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 325     /*     #endif */
 326     /* #endif */
 327
 328     /* Avoid stupid compiler warnings */
 329     jnrA = jnrB = jnrC = jnrD = 0;
 330     j_coord_offsetA = 0;
 331     j_coord_offsetB = 0;
 332     j_coord_offsetC = 0;
 333     j_coord_offsetD = 0;
 334
 335     /* ## Keep track of the floating point operations we issue for reporting! */
 336     /* #define OUTERFLOPS 0 */
 337     outeriter        = 0;
 338     inneriter        = 0;
 339
 340     for(iidx=0;iidx<4*DIM;iidx++)
 341     {
 342         scratch[iidx] = 0.0;
 343     }
 344
 345     /* Start outer loop over neighborlists */
 346     for(iidx=0; iidx<nri; iidx++)
 347     {
 348         /* Load shift vector for this list */
 349         i_shift_offset   = DIM*shiftidx[iidx];
 350
 351         /* Load limits for loop over neighbors */
 352         j_index_start    = jindex[iidx];
 353         j_index_end      = jindex[iidx+1];
 354
 355         /* Get outer coordinate index */
 356         inr              = iinr[iidx];
 357         i_coord_offset   = DIM*inr;
 358
 359         /* Load i particle coords and add shift vector */
 360         /* #if GEOMETRY_I == 'Particle' */
 361         gmx_mm256_load_shift_and_1rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 362         /* #elif GEOMETRY_I == 'Water3' */
 363         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 364                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 365         /* #elif GEOMETRY_I == 'Water4' */
 366         /*     #if 0 in PARTICLES_I                 */
 367         gmx_mm256_load_shift_and_4rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 368                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 369         /*     #else                                */
 370         gmx_mm256_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 371                                                     &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 372         /*     #endif                               */
 373         /* #endif                                   */
 374
 375         /* #if 'Force' in KERNEL_VF */
 376         /*     #for I in PARTICLES_I */
 377         fix{I}             = _mm256_setzero_pd();
 378         fiy{I}             = _mm256_setzero_pd();
 379         fiz{I}             = _mm256_setzero_pd();
 380         /*     #endfor */
 381         /* #endif */
 382
 383         /* ## For water we already preloaded parameters at the start of the kernel */
 384         /* #if not 'Water' in GEOMETRY_I */
 385         /* Load parameters for i particles */
 386         /*     #for I in PARTICLES_ELEC_I */
 387         iq{I}              = _mm256_mul_pd(facel,_mm256_set1_pd(charge[inr+{I}]));
 388         /*         #define OUTERFLOPS OUTERFLOPS+1 */
 389         /*         #if KERNEL_ELEC=='GeneralizedBorn' */
 390         isai{I}            = _mm256_set1_pd(invsqrta[inr+{I}]);
 391         /*         #endif */
 392         /*     #endfor */
 393         /*     #for I in PARTICLES_VDW_I */
 394         vdwioffsetptr{I}   = vdwparam+2*nvdwtype*vdwtype[inr+{I}];
 395         /*         #if 'LJEwald' in KERNEL_VDW */
 396         vdwgridioffsetptr{I} = vdwgridparam+2*nvdwtype*vdwtype[inr+{I}];
 397         /*         #endif */
 398         /*     #endfor */
 399         /* #endif */
 400
 401         /* #if 'Potential' in KERNEL_VF */
 402         /* Reset potential sums */
 403         /*     #if KERNEL_ELEC != 'None' */
 404         velecsum         = _mm256_setzero_pd();
 405         /*     #endif */
 406         /*     #if 'GeneralizedBorn' in KERNEL_ELEC */
 407         vgbsum           = _mm256_setzero_pd();
 408         /*     #endif */
 409         /*     #if KERNEL_VDW != 'None' */
 410         vvdwsum          = _mm256_setzero_pd();
 411         /*     #endif */
 412         /* #endif */
 413         /*     #if 'GeneralizedBorn' in KERNEL_ELEC and 'Force' in KERNEL_VF */
 414         dvdasum          = _mm256_setzero_pd();
 415         /*     #endif */
 416
 417         /* #for ROUND in ['Loop','Epilogue'] */
 418
 419         /* #if ROUND =='Loop' */
 420         /* Start inner kernel loop */
 421         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 422         {
 423         /* ## First round is normal loop (next statement resets indentation) */
 424         /*     #if 0 */
 425         }
 426         /*     #endif */
 427         /* #else */
 428         if(jidx<j_index_end)
 429         {
 430         /* ## Second round is epilogue */
 431         /* #endif */
 432         /* #define INNERFLOPS 0 */
 433
 434             /* Get j neighbor index, and coordinate index */
 435             /* #if ROUND =='Loop' */
 436             jnrA             = jjnr[jidx];
 437             jnrB             = jjnr[jidx+1];
 438             jnrC             = jjnr[jidx+2];
 439             jnrD             = jjnr[jidx+3];
 440             /* #else */
 441             jnrlistA         = jjnr[jidx];
 442             jnrlistB         = jjnr[jidx+1];
 443             jnrlistC         = jjnr[jidx+2];
 444             jnrlistD         = jjnr[jidx+3];
 445             /* Sign of each element will be negative for non-real atoms.
 446              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 447              * so use it as val = _mm_andnot_pd(mask,val) to clear dummy entries.
 448              */
 449             tmpmask0 = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 450
 451             tmpmask1 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(3,3,2,2));
 452             tmpmask0 = _mm_permute_ps(tmpmask0,_GMX_MM_PERMUTE(1,1,0,0));
 453             dummy_mask = _mm256_castps_pd(gmx_mm256_set_m128(tmpmask1,tmpmask0));
 454
 455             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 456             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 457             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 458             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 459             /* #endif */
 460             j_coord_offsetA  = DIM*jnrA;
 461             j_coord_offsetB  = DIM*jnrB;
 462             j_coord_offsetC  = DIM*jnrC;
 463             j_coord_offsetD  = DIM*jnrD;
 464
 465             /* load j atom coordinates */
 466             /* #if GEOMETRY_J == 'Particle'             */
 467             gmx_mm256_load_1rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 468                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 469                                                  &jx0,&jy0,&jz0);
 470             /* #elif GEOMETRY_J == 'Water3'             */
 471             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 472                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 473                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 474             /* #elif GEOMETRY_J == 'Water4'             */
 475             /*     #if 0 in PARTICLES_J                 */
 476             gmx_mm256_load_4rvec_4ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 477                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 478                                                  &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,
 479                                                  &jy2,&jz2,&jx3,&jy3,&jz3);
 480             /*     #else                                */
 481             gmx_mm256_load_3rvec_4ptr_swizzle_pd(x+j_coord_offsetA+DIM,x+j_coord_offsetB+DIM,
 482                                                  x+j_coord_offsetC+DIM,x+j_coord_offsetD+DIM,
 483                                                  &jx1,&jy1,&jz1,&jx2,&jy2,&jz2,&jx3,&jy3,&jz3);
 484             /*     #endif                               */
 485             /* #endif                                   */
 486
 487             /* Calculate displacement vector */
 488             /* #for I,J in PAIRS_IJ */
 489             dx{I}{J}             = _mm256_sub_pd(ix{I},jx{J});
 490             dy{I}{J}             = _mm256_sub_pd(iy{I},jy{J});
 491             dz{I}{J}             = _mm256_sub_pd(iz{I},jz{J});
 492             /*     #define INNERFLOPS INNERFLOPS+3 */
 493             /* #endfor */
 494
 495             /* Calculate squared distance and things based on it */
 496             /* #for I,J in PAIRS_IJ */
 497             rsq{I}{J}            = gmx_mm256_calc_rsq_pd(dx{I}{J},dy{I}{J},dz{I}{J});
 498             /*     #define INNERFLOPS INNERFLOPS+5 */
 499             /* #endfor */
 500
 501             /* #for I,J in PAIRS_IJ */
 502             /*     #if 'rinv' in INTERACTION_FLAGS[I][J] */
 503             rinv{I}{J}           = gmx_mm256_invsqrt_pd(rsq{I}{J});
 504             /*         #define INNERFLOPS INNERFLOPS+5 */
 505             /*     #endif */
 506             /* #endfor */
 507
 508             /* #for I,J in PAIRS_IJ */
 509             /*     #if 'rinvsq' in INTERACTION_FLAGS[I][J] */
 510             /*         # if 'rinv' not in INTERACTION_FLAGS[I][J] */
 511             rinvsq{I}{J}         = gmx_mm256_inv_pd(rsq{I}{J});
 512             /*             #define INNERFLOPS INNERFLOPS+4 */
 513             /*         #else */
 514             rinvsq{I}{J}         = _mm256_mul_pd(rinv{I}{J},rinv{I}{J});
 515             /*             #define INNERFLOPS INNERFLOPS+1 */
 516             /*         #endif */
 517             /*     #endif */
 518             /* #endfor */
 519
 520             /* #if not 'Water' in GEOMETRY_J */
 521             /* Load parameters for j particles */
 522             /*     #for J in PARTICLES_ELEC_J */
 523             jq{J}              = gmx_mm256_load_4real_swizzle_pd(charge+jnrA+{J},charge+jnrB+{J},
 524                                                                  charge+jnrC+{J},charge+jnrD+{J});
 525             /*         #if KERNEL_ELEC=='GeneralizedBorn' */
 526             isaj{J}            = gmx_mm256_load_4real_swizzle_pd(invsqrta+jnrA+{J},invsqrta+jnrB+{J},
 527                                                                  invsqrta+jnrC+{J},invsqrta+jnrD+{J});
 528             /*         #endif */
 529             /*     #endfor */
 530             /*     #for J in PARTICLES_VDW_J */
 531             vdwjidx{J}A        = 2*vdwtype[jnrA+{J}];
 532             vdwjidx{J}B        = 2*vdwtype[jnrB+{J}];
 533             vdwjidx{J}C        = 2*vdwtype[jnrC+{J}];
 534             vdwjidx{J}D        = 2*vdwtype[jnrD+{J}];
 535             /*     #endfor */
 536             /* #endif */
 537
 538             /* #if 'Force' in KERNEL_VF and not 'Particle' in GEOMETRY_I */
 539             /*     #for J in PARTICLES_J */
 540             fjx{J}             = _mm256_setzero_pd();
 541             fjy{J}             = _mm256_setzero_pd();
 542             fjz{J}             = _mm256_setzero_pd();
 543             /*     #endfor */
 544             /* #endif */
 545
 546             /* #for I,J in PAIRS_IJ */
 547
 548             /**************************
 549              * CALCULATE INTERACTIONS *
 550              **************************/
 551
 552             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 553             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 554             /*         ## We always calculate rinv/rinvsq above to enable pipelineing in compilers (performance tested on x86) */
 555             if (gmx_mm256_any_lt(rsq{I}{J},rcutoff2))
 556             {
 557                 /*     #if 0    ## this and the next two lines is a hack to maintain auto-indentation in template file */
 558             }
 559             /*         #endif */
 560             /*         #define INNERFLOPS INNERFLOPS+1 */
 561             /*     #endif */
 562
 563             /*     #if 'r' in INTERACTION_FLAGS[I][J] */
 564             r{I}{J}              = _mm256_mul_pd(rsq{I}{J},rinv{I}{J});
 565             /*         #if ROUND == 'Epilogue' */
 566             r{I}{J}              = _mm256_andnot_pd(dummy_mask,r{I}{J});
 567             /*             #define INNERFLOPS INNERFLOPS+1 */
 568             /*         #endif */
 569             /*         #define INNERFLOPS INNERFLOPS+1 */
 570             /*     #endif */
 571
 572             /*     ## For water geometries we already loaded parameters at the start of the kernel */
 573             /*     #if not 'Water' in GEOMETRY_J */
 574             /* Compute parameters for interactions between i and j atoms */
 575             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 576             qq{I}{J}             = _mm256_mul_pd(iq{I},jq{J});
 577             /*             #define INNERFLOPS INNERFLOPS+1 */
 578             /*         #endif */
 579             /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 580             gmx_mm256_load_4pair_swizzle_pd(vdwioffsetptr{I}+vdwjidx{J}A,
 581                                             vdwioffsetptr{I}+vdwjidx{J}B,
 582                                             vdwioffsetptr{I}+vdwjidx{J}C,
 583                                             vdwioffsetptr{I}+vdwjidx{J}D,
 584                                             &c6_{I}{J},&c12_{I}{J});
 585
 586             /*           #if 'LJEwald' in KERNEL_VDW */
 587             c6grid_{I}{J}       = gmx_mm256_load_4real_swizzle_pd(vdwgridioffsetptr{I}+vdwjidx{J}A,
 588                                                                   vdwgridioffsetptr{I}+vdwjidx{J}B,
 589                                                                   vdwgridioffsetptr{I}+vdwjidx{J}C,
 590                                                                   vdwgridioffsetptr{I}+vdwjidx{J}D);
 591             /*           #endif */
 592             /*         #endif */
 593             /*     #endif */
 594
 595             /*     #if 'table' in INTERACTION_FLAGS[I][J] */
 596             /* Calculate table index by multiplying r with table scale and truncate to integer */
 597             rt               = _mm256_mul_pd(r{I}{J},vftabscale);
 598             vfitab           = _mm256_cvttpd_epi32(rt);
 599             vfeps            = _mm256_sub_pd(rt,_mm256_round_pd(rt, _MM_FROUND_FLOOR));
 600             /*         #define INNERFLOPS INNERFLOPS+4                          */
 601             /*         #if 'Table' in KERNEL_ELEC and 'Table' in KERNEL_VDW     */
 602             /*             ## 3 tables, 4 bytes per point: multiply index by 12 */
 603             vfitab           = _mm_slli_epi32(_mm_add_epi32(vfitab,_mm_slli_epi32(vfitab,1)),2);
 604             /*         #elif 'Table' in KERNEL_ELEC                             */
 605             /*             ## 1 table, 4 bytes per point: multiply index by 4   */
 606             vfitab           = _mm_slli_epi32(vfitab,2);
 607             /*         #elif 'Table' in KERNEL_VDW                              */
 608             /*             ## 2 tables, 4 bytes per point: multiply index by 8  */
 609             vfitab           = _mm_slli_epi32(vfitab,3);
 610             /*         #endif                                                   */
 611             /*     #endif */
 612
 613             /*     ## ELECTROSTATIC INTERACTIONS */
 614             /*     #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 615
 616             /*         #if KERNEL_ELEC=='Coulomb' */
 617
 618             /* COULOMB ELECTROSTATICS */
 619             velec            = _mm256_mul_pd(qq{I}{J},rinv{I}{J});
 620             /*             #define INNERFLOPS INNERFLOPS+1 */
 621             /*             #if 'Force' in KERNEL_VF */
 622             felec            = _mm256_mul_pd(velec,rinvsq{I}{J});
 623             /*                 #define INNERFLOPS INNERFLOPS+1 */
 624             /*             #endif */
 625
 626             /*         #elif KERNEL_ELEC=='ReactionField' */
 627
 628             /* REACTION-FIELD ELECTROSTATICS */
 629             /*             #if 'Potential' in KERNEL_VF */
 630             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_add_pd(rinv{I}{J},_mm256_mul_pd(krf,rsq{I}{J})),crf));
 631             /*                 #define INNERFLOPS INNERFLOPS+4 */
 632             /*             #endif */
 633             /*             #if 'Force' in KERNEL_VF */
 634             felec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_mul_pd(rinv{I}{J},rinvsq{I}{J}),krf2));
 635             /*                 #define INNERFLOPS INNERFLOPS+3 */
 636             /*             #endif */
 637
 638             /*         #elif KERNEL_ELEC=='GeneralizedBorn' */
 639
 640             /* GENERALIZED BORN AND COULOMB ELECTROSTATICS */
 641             isaprod          = _mm256_mul_pd(isai{I},isaj{J});
 642             gbqqfactor       = _mm256_xor_pd(signbit,_mm256_mul_pd(qq{I}{J},_mm256_mul_pd(isaprod,gbinvepsdiff)));
 643             gbscale          = _mm256_mul_pd(isaprod,gbtabscale);
 644             /*             #define INNERFLOPS INNERFLOPS+5 */
 645
 646             /* Calculate generalized born table index - this is a separate table from the normal one,
 647              * but we use the same procedure by multiplying r with scale and truncating to integer.
 648              */
 649             rt               = _mm256_mul_pd(r{I}{J},gbscale);
 650             gbitab           = _mm256_cvttpd_epi32(rt);
 651             gbeps            = _mm256_sub_pd(rt,_mm256_round_pd(rt, _MM_FROUND_FLOOR));
 652             gbitab           = _mm_slli_epi32(gbitab,2);
 653             Y                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,0) );
 654             F                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,1) );
 655             G                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,2) );
 656             H                = _mm256_load_pd( gbtab + _mm_extract_epi32(gbitab,3) );
 657             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 658             Heps             = _mm256_mul_pd(gbeps,H);
 659             Fp               = _mm256_add_pd(F,_mm256_mul_pd(gbeps,_mm256_add_pd(G,Heps)));
 660             VV               = _mm256_add_pd(Y,_mm256_mul_pd(gbeps,Fp));
 661             vgb              = _mm256_mul_pd(gbqqfactor,VV);
 662             /*             #define INNERFLOPS INNERFLOPS+10 */
 663
 664             /*             #if 'Force' in KERNEL_VF */
 665             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 666             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
 667             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r{I}{J})));
 668             /*                 #if ROUND == 'Epilogue' */
 669             dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
 670             /*                 #endif */
 671             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
 672             /*                 #if ROUND == 'Loop' */
 673             fjptrA           = dvda+jnrA;
 674             fjptrB           = dvda+jnrB;
 675             fjptrC           = dvda+jnrC;
 676             fjptrD           = dvda+jnrD;
 677             /*                 #else */
 678             /* 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. */
 679             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;
 680             fjptrB             = (jnrlistB>=0) ? dvda+jnrB : scratch;
 681             fjptrC             = (jnrlistC>=0) ? dvda+jnrC : scratch;
 682             fjptrD             = (jnrlistD>=0) ? dvda+jnrD : scratch;
 683             /*                 #endif */
 684             gmx_mm256_increment_4real_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
 685                                                  _mm256_mul_pd(dvdatmp,_mm256_mul_pd(isaj{J},isaj{J})));
 686             /*                 #define INNERFLOPS INNERFLOPS+12 */
 687             /*             #endif */
 688             velec            = _mm256_mul_pd(qq{I}{J},rinv{I}{J});
 689             /*                 #define INNERFLOPS INNERFLOPS+1 */
 690             /*             #if 'Force' in KERNEL_VF */
 691             felec            = _mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(velec,rinv{I}{J}),fgb),rinv{I}{J});
 692             /*                 #define INNERFLOPS INNERFLOPS+3 */
 693             /*             #endif */
 694
 695             /*         #elif KERNEL_ELEC=='Ewald' */
 696             /* EWALD ELECTROSTATICS */
 697
 698             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 699             ewrt             = _mm256_mul_pd(r{I}{J},ewtabscale);
 700             ewitab           = _mm256_cvttpd_epi32(ewrt);
 701             eweps            = _mm256_sub_pd(ewrt,_mm256_round_pd(ewrt, _MM_FROUND_FLOOR));
 702             /*             #define INNERFLOPS INNERFLOPS+4 */
 703             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_ELEC=='PotentialSwitch' */
 704             ewitab           = _mm_slli_epi32(ewitab,2);
 705             ewtabF           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,0) );
 706             ewtabD           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,1) );
 707             ewtabV           = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,2) );
 708             ewtabFn          = _mm256_load_pd( ewtab + _mm_extract_epi32(ewitab,3) );
 709             GMX_MM256_FULLTRANSPOSE4_PD(ewtabF,ewtabD,ewtabV,ewtabFn);
 710             felec            = _mm256_add_pd(ewtabF,_mm256_mul_pd(eweps,ewtabD));
 711             /*                 #define INNERFLOPS INNERFLOPS+2 */
 712             /*                 #if KERNEL_MOD_ELEC=='PotentialShift' */
 713             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 714             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(_mm256_sub_pd(rinv{I}{J},sh_ewald),velec));
 715             /*                     #define INNERFLOPS INNERFLOPS+7 */
 716             /*                 #else */
 717             velec            = _mm256_sub_pd(ewtabV,_mm256_mul_pd(_mm256_mul_pd(ewtabhalfspace,eweps),_mm256_add_pd(ewtabF,felec)));
 718             velec            = _mm256_mul_pd(qq{I}{J},_mm256_sub_pd(rinv{I}{J},velec));
 719             /*                     #define INNERFLOPS INNERFLOPS+6 */
 720             /*                 #endif */
 721             /*                 #if 'Force' in KERNEL_VF */
 722             felec            = _mm256_mul_pd(_mm256_mul_pd(qq{I}{J},rinv{I}{J}),_mm256_sub_pd(rinvsq{I}{J},felec));
 723             /*                      #define INNERFLOPS INNERFLOPS+3 */
 724             /*                 #endif */
 725             /*             #elif KERNEL_VF=='Force' */
 726             gmx_mm256_load_4pair_swizzle_pd(ewtab + _mm_extract_epi32(ewitab,0),ewtab + _mm_extract_epi32(ewitab,1),
 727                                             ewtab + _mm_extract_epi32(ewitab,2),ewtab + _mm_extract_epi32(ewitab,3),
 728                                             &ewtabF,&ewtabFn);
 729             felec            = _mm256_add_pd(_mm256_mul_pd( _mm256_sub_pd(one,eweps),ewtabF),_mm256_mul_pd(eweps,ewtabFn));
 730             felec            = _mm256_mul_pd(_mm256_mul_pd(qq{I}{J},rinv{I}{J}),_mm256_sub_pd(rinvsq{I}{J},felec));
 731             /*                 #define INNERFLOPS INNERFLOPS+7 */
 732             /*             #endif */
 733
 734             /*         #elif KERNEL_ELEC=='CubicSplineTable' */
 735
 736             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 737             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 738             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 739             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 740             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 741             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 742             Heps             = _mm256_mul_pd(vfeps,H);
 743             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 744             /*             #define INNERFLOPS INNERFLOPS+4 */
 745             /*             #if 'Potential' in KERNEL_VF */
 746             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 747             velec            = _mm256_mul_pd(qq{I}{J},VV);
 748             /*                 #define INNERFLOPS INNERFLOPS+3 */
 749             /*             #endif */
 750             /*             #if 'Force' in KERNEL_VF */
 751             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 752             felec            = _mm256_xor_pd(signbit,_mm256_mul_pd(_mm256_mul_pd(qq{I}{J},FF),_mm256_mul_pd(vftabscale,rinv{I}{J})));
 753             /*                 #define INNERFLOPS INNERFLOPS+7 */
 754             /*             #endif */
 755             /*         #endif */
 756             /*         ## End of check for electrostatics interaction forms */
 757             /*     #endif */
 758             /*     ## END OF ELECTROSTATIC INTERACTION CHECK FOR PAIR I-J */
 759
 760             /*     #if 'vdw' in INTERACTION_FLAGS[I][J] */
 761
 762             /*         #if KERNEL_VDW=='LennardJones' */
 763
 764             /* LENNARD-JONES DISPERSION/REPULSION */
 765
 766             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq{I}{J},rinvsq{I}{J}),rinvsq{I}{J});
 767             /*             #define INNERFLOPS INNERFLOPS+2 */
 768             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_VDW=='PotentialSwitch' */
 769             vvdw6            = _mm256_mul_pd(c6_{I}{J},rinvsix);
 770             vvdw12           = _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(rinvsix,rinvsix));
 771             /*                 #define INNERFLOPS INNERFLOPS+3 */
 772             /*                 #if KERNEL_MOD_VDW=='PotentialShift' */
 773             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) ,
 774                                           _mm256_mul_pd( _mm256_sub_pd(vvdw6,_mm256_mul_pd(c6_{I}{J},sh_vdw_invrcut6)),one_sixth));
 775             /*                     #define INNERFLOPS INNERFLOPS+8 */
 776             /*                 #else */
 777             vvdw             = _mm256_sub_pd( _mm256_mul_pd(vvdw12,one_twelfth) , _mm256_mul_pd(vvdw6,one_sixth) );
 778             /*                     #define INNERFLOPS INNERFLOPS+3 */
 779             /*                 #endif */
 780             /*                 ## Check for force inside potential check, i.e. this means we already did the potential part */
 781             /*                 #if 'Force' in KERNEL_VF */
 782             fvdw             = _mm256_mul_pd(_mm256_sub_pd(vvdw12,vvdw6),rinvsq{I}{J});
 783             /*                     #define INNERFLOPS INNERFLOPS+2 */
 784             /*                 #endif */
 785             /*             #elif KERNEL_VF=='Force' */
 786             /*                 ## Force-only LennardJones makes it possible to save 1 flop (they do add up...) */
 787             fvdw             = _mm256_mul_pd(_mm256_sub_pd(_mm256_mul_pd(c12_{I}{J},rinvsix),c6_{I}{J}),_mm256_mul_pd(rinvsix,rinvsq{I}{J}));
 788             /*                 #define INNERFLOPS INNERFLOPS+4 */
 789             /*             #endif */
 790
 791             /*         #elif KERNEL_VDW=='CubicSplineTable' */
 792
 793             /* CUBIC SPLINE TABLE DISPERSION */
 794             /*             #if 'Table' in KERNEL_ELEC */
 795             vfitab           = _mm_add_epi32(vfitab,ifour);
 796             /*             #endif                     */
 797             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 798             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 799             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 800             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 801             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 802             Heps             = _mm256_mul_pd(vfeps,H);
 803             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 804             /*             #define INNERFLOPS INNERFLOPS+4 */
 805             /*             #if 'Potential' in KERNEL_VF */
 806             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 807             vvdw6            = _mm256_mul_pd(c6_{I}{J},VV);
 808             /*                 #define INNERFLOPS INNERFLOPS+3 */
 809             /*             #endif */
 810             /*             #if 'Force' in KERNEL_VF */
 811             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 812             fvdw6            = _mm256_mul_pd(c6_{I}{J},FF);
 813             /*                 #define INNERFLOPS INNERFLOPS+4 */
 814             /*             #endif */
 815
 816             /* CUBIC SPLINE TABLE REPULSION */
 817             vfitab           = _mm_add_epi32(vfitab,ifour);
 818             Y                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 819             F                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 820             G                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,2) );
 821             H                = _mm256_load_pd( vftab + _mm_extract_epi32(vfitab,3) );
 822             GMX_MM256_FULLTRANSPOSE4_PD(Y,F,G,H);
 823             Heps             = _mm256_mul_pd(vfeps,H);
 824             Fp               = _mm256_add_pd(F,_mm256_mul_pd(vfeps,_mm256_add_pd(G,Heps)));
 825             /*             #define INNERFLOPS INNERFLOPS+4 */
 826             /*             #if 'Potential' in KERNEL_VF */
 827             VV               = _mm256_add_pd(Y,_mm256_mul_pd(vfeps,Fp));
 828             vvdw12           = _mm256_mul_pd(c12_{I}{J},VV);
 829             /*                 #define INNERFLOPS INNERFLOPS+3 */
 830             /*             #endif */
 831             /*             #if 'Force' in KERNEL_VF */
 832             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(vfeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
 833             fvdw12           = _mm256_mul_pd(c12_{I}{J},FF);
 834             /*                 #define INNERFLOPS INNERFLOPS+5 */
 835             /*             #endif */
 836             /*             #if 'Potential' in KERNEL_VF */
 837             vvdw             = _mm256_add_pd(vvdw12,vvdw6);
 838             /*                 #define INNERFLOPS INNERFLOPS+1 */
 839             /*             #endif */
 840             /*             #if 'Force' in KERNEL_VF */
 841             fvdw             = _mm256_xor_pd(signbit,_mm256_mul_pd(_mm256_add_pd(fvdw6,fvdw12),_mm256_mul_pd(vftabscale,rinv{I}{J})));
 842             /*                 #define INNERFLOPS INNERFLOPS+4 */
 843             /*             #endif */
 844
 845             /*         #elif KERNEL_VDW=='LJEwald' */
 846
 847             /* Analytical LJ-PME */
 848             rinvsix          = _mm256_mul_pd(_mm256_mul_pd(rinvsq{I}{J},rinvsq{I}{J}),rinvsq{I}{J});
 849             ewcljrsq         = _mm256_mul_pd(ewclj2,rsq{I}{J});
 850             ewclj6           = _mm256_mul_pd(ewclj2,_mm256_mul_pd(ewclj2,ewclj2));
 851             exponent         = gmx_simd_exp_d(ewcljrsq);
 852             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 853             poly             = _mm256_mul_pd(exponent,_mm256_add_pd(_mm256_sub_pd(one,ewcljrsq),_mm256_mul_pd(_mm256_mul_pd(ewcljrsq,ewcljrsq),one_half)));
 854             /*                 #define INNERFLOPS INNERFLOPS+11 */
 855             /*             #if 'Potential' in KERNEL_VF or KERNEL_MOD_VDW=='PotentialSwitch' */
 856             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 857             vvdw6            = _mm256_mul_pd(_mm256_sub_pd(c6_{I}{J},_mm256_mul_pd(c6grid_{I}{J},_mm256_sub_pd(one,poly))),rinvsix);
 858             vvdw12           = _mm256_mul_pd(c12_{I}{J},_mm256_mul_pd(rinvsix,rinvsix));
 859             /*                     #define INNERFLOPS INNERFLOPS+6 */
 860             /*                 #if KERNEL_MOD_VDW=='PotentialShift' */
 861             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) ,
 862                                           _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));
 863             /*                     #define INNERFLOPS INNERFLOPS+10 */
 864             /*                 #else */
 865             vvdw             = _mm256_sub_pd(_mm256_mul_pd(vvdw12,one_twelfth),_mm256_mul_pd(vvdw6,one_sixth));
 866             /*                 #define INNERFLOPS INNERFLOPS+6 */
 867             /*                 #endif */
 868             /*                  ## Check for force inside potential check, i.e. this means we already did the potential part */
 869             /*                  #if 'Force' in KERNEL_VF */
 870             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 871             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});
 872             /*                 #define INNERFLOPS INNERFLOPS+6 */
 873             /*                  #endif */
 874             /*              #elif KERNEL_VF=='Force' */
 875             /* f6A = 6 * C6grid * (1 - poly) */
 876             f6A              = _mm256_mul_pd(c6grid_{I}{J},_mm256_sub_pd(one,poly));
 877             /* f6B = C6grid * exponent * beta^6 */
 878             f6B              = _mm256_mul_pd(_mm256_mul_pd(c6grid_{I}{J},one_sixth),_mm256_mul_pd(exponent,ewclj6));
 879             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 880             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});
 881             /*                 #define INNERFLOPS INNERFLOPS+11 */
 882             /*              #endif */
 883             /*         #endif */
 884             /*         ## End of check for vdw interaction forms */
 885             /*     #endif */
 886             /*     ## END OF VDW INTERACTION CHECK FOR PAIR I-J */
 887
 888             /*     #if 'switch' in INTERACTION_FLAGS[I][J] */
 889             d                = _mm256_sub_pd(r{I}{J},rswitch);
 890             d                = _mm256_max_pd(d,_mm256_setzero_pd());
 891             d2               = _mm256_mul_pd(d,d);
 892             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)))))));
 893             /*         #define INNERFLOPS INNERFLOPS+10 */
 894
 895             /*         #if 'Force' in KERNEL_VF */
 896             dsw              = _mm256_mul_pd(d2,_mm256_add_pd(swF2,_mm256_mul_pd(d,_mm256_add_pd(swF3,_mm256_mul_pd(d,swF4)))));
 897             /*             #define INNERFLOPS INNERFLOPS+5 */
 898             /*         #endif */
 899
 900             /* Evaluate switch function */
 901             /*         #if 'Force' in KERNEL_VF */
 902             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 903             /*             #if 'electrostatics' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_ELEC=='PotentialSwitch' */
 904             felec            = _mm256_sub_pd( _mm256_mul_pd(felec,sw) , _mm256_mul_pd(rinv{I}{J},_mm256_mul_pd(velec,dsw)) );
 905             /*                 #define INNERFLOPS INNERFLOPS+4 */
 906             /*             #endif */
 907             /*             #if 'vdw' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_VDW=='PotentialSwitch' */
 908             fvdw             = _mm256_sub_pd( _mm256_mul_pd(fvdw,sw) , _mm256_mul_pd(rinv{I}{J},_mm256_mul_pd(vvdw,dsw)) );
 909             /*                 #define INNERFLOPS INNERFLOPS+4 */
 910             /*             #endif */
 911             /*         #endif */
 912             /*         #if 'Potential' in KERNEL_VF */
 913             /*             #if 'electrostatics' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_ELEC=='PotentialSwitch' */
 914             velec            = _mm256_mul_pd(velec,sw);
 915             /*                 #define INNERFLOPS INNERFLOPS+1 */
 916             /*             #endif */
 917             /*             #if 'vdw' in INTERACTION_FLAGS[I][J] and KERNEL_MOD_VDW=='PotentialSwitch' */
 918             vvdw             = _mm256_mul_pd(vvdw,sw);
 919             /*                 #define INNERFLOPS INNERFLOPS+1 */
 920             /*             #endif */
 921             /*         #endif */
 922             /*     #endif */
 923             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 924             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 925             cutoff_mask      = _mm256_cmp_pd(rsq{I}{J},rcutoff2,_CMP_LT_OQ);
 926             /*         #define INNERFLOPS INNERFLOPS+1 */
 927             /*     #endif */
 928
 929             /*     #if 'Potential' in KERNEL_VF */
 930             /* Update potential sum for this i atom from the interaction with this j atom. */
 931             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] */
 932             /*             #if 'exactcutoff' in INTERACTION_FLAGS[I][J] */
 933             velec            = _mm256_and_pd(velec,cutoff_mask);
 934             /*                 #define INNERFLOPS INNERFLOPS+1 */
 935             /*             #endif                                       */
 936             /*             #if ROUND == 'Epilogue' */
 937             velec            = _mm256_andnot_pd(dummy_mask,velec);
 938             /*             #endif */
 939             velecsum         = _mm256_add_pd(velecsum,velec);
 940             /*             #define INNERFLOPS INNERFLOPS+1 */
 941             /*             #if KERNEL_ELEC=='GeneralizedBorn' */
 942             /*             #if 'exactcutoff' in INTERACTION_FLAGS[I][J] */
 943             vgb              = _mm256_and_pd(vgb,cutoff_mask);
 944             /*                 #define INNERFLOPS INNERFLOPS+1 */
 945             /*             #endif                                       */
 946             /*             #if ROUND == 'Epilogue' */
 947             vgb              = _mm256_andnot_pd(dummy_mask,vgb);
 948             /*             #endif */
 949             vgbsum           = _mm256_add_pd(vgbsum,vgb);
 950             /*                 #define INNERFLOPS INNERFLOPS+1 */
 951             /*             #endif */
 952             /*         #endif */
 953             /*         #if 'vdw' in INTERACTION_FLAGS[I][J] */
 954             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 955             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 956             vvdw             = _mm256_and_pd(vvdw,cutoff_mask);
 957             /*                 #define INNERFLOPS INNERFLOPS+1 */
 958             /*             #endif                                       */
 959             /*             #if ROUND == 'Epilogue' */
 960             vvdw             = _mm256_andnot_pd(dummy_mask,vvdw);
 961             /*             #endif */
 962             vvdwsum          = _mm256_add_pd(vvdwsum,vvdw);
 963             /*             #define INNERFLOPS INNERFLOPS+1 */
 964             /*         #endif */
 965             /*     #endif */
 966
 967             /*     #if 'Force' in KERNEL_VF */
 968
 969             /*         #if 'electrostatics' in INTERACTION_FLAGS[I][J] and 'vdw' in INTERACTION_FLAGS[I][J] */
 970             fscal            = _mm256_add_pd(felec,fvdw);
 971             /*             #define INNERFLOPS INNERFLOPS+1 */
 972             /*         #elif 'electrostatics' in INTERACTION_FLAGS[I][J] */
 973             fscal            = felec;
 974             /*         #elif 'vdw' in INTERACTION_FLAGS[I][J] */
 975             fscal            = fvdw;
 976             /*        #endif */
 977
 978             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
 979             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
 980             fscal            = _mm256_and_pd(fscal,cutoff_mask);
 981             /*                 #define INNERFLOPS INNERFLOPS+1 */
 982             /*             #endif                                       */
 983
 984             /*             #if ROUND == 'Epilogue' */
 985             fscal            = _mm256_andnot_pd(dummy_mask,fscal);
 986             /*             #endif */
 987
 988             /* Calculate temporary vectorial force */
 989             tx               = _mm256_mul_pd(fscal,dx{I}{J});
 990             ty               = _mm256_mul_pd(fscal,dy{I}{J});
 991             tz               = _mm256_mul_pd(fscal,dz{I}{J});
 992
 993             /* Update vectorial force */
 994             fix{I}             = _mm256_add_pd(fix{I},tx);
 995             fiy{I}             = _mm256_add_pd(fiy{I},ty);
 996             fiz{I}             = _mm256_add_pd(fiz{I},tz);
 997             /*             #define INNERFLOPS INNERFLOPS+6 */
 998
 999             /* #if GEOMETRY_I == 'Particle'             */
1000             /*     #if ROUND == 'Loop' */
1001             fjptrA             = f+j_coord_offsetA;
1002             fjptrB             = f+j_coord_offsetB;
1003             fjptrC             = f+j_coord_offsetC;
1004             fjptrD             = f+j_coord_offsetD;
1005             /*     #else */
1006             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1007             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1008             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1009             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1010             /*     #endif */
1011             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,tx,ty,tz);
1012             /*     #define INNERFLOPS INNERFLOPS+3      */
1013             /* #else                                    */
1014             fjx{J}             = _mm256_add_pd(fjx{J},tx);
1015             fjy{J}             = _mm256_add_pd(fjy{J},ty);
1016             fjz{J}             = _mm256_add_pd(fjz{J},tz);
1017             /*     #define INNERFLOPS INNERFLOPS+3      */
1018             /* #endif                                   */
1019
1020             /*     #endif */
1021
1022             /*     ## Note special check for TIP4P-TIP4P. Since we are cutting of all hydrogen interactions we also cut the LJ-only O-O interaction */
1023             /*     #if 'exactcutoff' in INTERACTION_FLAGS[I][J] or (GEOMETRY_I=='Water4' and GEOMETRY_J=='Water4' and 'exactcutoff' in INTERACTION_FLAGS[1][1]) */
1024             /*         #if 0    ## This and next two lines is a hack to maintain indentation in template file */
1025             {
1026                 /*     #endif */
1027             }
1028             /*     #endif */
1029             /*    ## End of check for the interaction being outside the cutoff */
1030
1031             /* #endfor */
1032             /* ## End of loop over i-j interaction pairs */
1033
1034             /* #if GEOMETRY_I != 'Particle' */
1035             /*     #if ROUND == 'Loop' */
1036             fjptrA             = f+j_coord_offsetA;
1037             fjptrB             = f+j_coord_offsetB;
1038             fjptrC             = f+j_coord_offsetC;
1039             fjptrD             = f+j_coord_offsetD;
1040             /*     #else */
1041             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1042             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1043             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1044             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1045             /*     #endif */
1046             /* #endif */
1047
1048             /* #if 'Water' in GEOMETRY_I and GEOMETRY_J == 'Particle' */
1049             gmx_mm256_decrement_1rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1050             /*     #define INNERFLOPS INNERFLOPS+3      */
1051             /* #elif GEOMETRY_J == 'Water3'             */
1052             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1053                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1054             /*     #define INNERFLOPS INNERFLOPS+9      */
1055             /* #elif GEOMETRY_J == 'Water4'             */
1056             /*     #if 0 in PARTICLES_J                 */
1057             gmx_mm256_decrement_4rvec_4ptr_swizzle_pd(fjptrA,fjptrB,fjptrC,fjptrD,
1058                                                       fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,
1059                                                       fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1060             /*     #define INNERFLOPS INNERFLOPS+12     */
1061             /*     #else                                */
1062             gmx_mm256_decrement_3rvec_4ptr_swizzle_pd(fjptrA+DIM,fjptrB+DIM,fjptrC+DIM,fjptrD+DIM,
1063                                                       fjx1,fjy1,fjz1,fjx2,fjy2,fjz2,fjx3,fjy3,fjz3);
1064             /*     #define INNERFLOPS INNERFLOPS+9      */
1065             /*     #endif                               */
1066             /* #endif                                   */
1067
1068             /* Inner loop uses {INNERFLOPS} flops */
1069         }
1070
1071         /* #endfor */
1072
1073         /* End of innermost loop */
1074
1075         /* #if 'Force' in KERNEL_VF */
1076         /*     #if GEOMETRY_I == 'Particle'            */
1077         gmx_mm256_update_iforce_1atom_swizzle_pd(fix0,fiy0,fiz0,
1078                                                  f+i_coord_offset,fshift+i_shift_offset);
1079         /*         #define OUTERFLOPS OUTERFLOPS+6     */
1080         /*     #elif GEOMETRY_I == 'Water3'            */
1081         gmx_mm256_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1082                                                  f+i_coord_offset,fshift+i_shift_offset);
1083         /*         #define OUTERFLOPS OUTERFLOPS+18    */
1084         /*     #elif GEOMETRY_I == 'Water4'            */
1085         /*         #if 0 in PARTICLES_I                */
1086         gmx_mm256_update_iforce_4atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1087                                                  f+i_coord_offset,fshift+i_shift_offset);
1088         /*             #define OUTERFLOPS OUTERFLOPS+24    */
1089         /*         #else                               */
1090         gmx_mm256_update_iforce_3atom_swizzle_pd(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1091                                                  f+i_coord_offset+DIM,fshift+i_shift_offset);
1092         /*             #define OUTERFLOPS OUTERFLOPS+18    */
1093         /*         #endif                              */
1094         /*     #endif                                  */
1095         /* #endif                                      */
1096
1097         /* #if 'Potential' in KERNEL_VF */
1098         ggid                        = gid[iidx];
1099         /* Update potential energies */
1100         /*     #if KERNEL_ELEC != 'None' */
1101         gmx_mm256_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
1102         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1103         /*     #endif */
1104         /*     #if 'GeneralizedBorn' in KERNEL_ELEC */
1105         gmx_mm256_update_1pot_pd(vgbsum,kernel_data->energygrp_polarization+ggid);
1106         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1107         /*     #endif */
1108         /*     #if KERNEL_VDW != 'None' */
1109         gmx_mm256_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
1110         /*         #define OUTERFLOPS OUTERFLOPS+1 */
1111         /*     #endif */
1112         /* #endif */
1113         /*     #if 'GeneralizedBorn' in KERNEL_ELEC and 'Force' in KERNEL_VF */
1114         dvdasum = _mm256_mul_pd(dvdasum, _mm256_mul_pd(isai{I},isai{I}));
1115         gmx_mm256_update_1pot_pd(dvdasum,dvda+inr);
1116         /*     #endif */
1117
1118         /* Increment number of inner iterations */
1119         inneriter                  += j_index_end - j_index_start;
1120
1121         /* Outer loop uses {OUTERFLOPS} flops */
1122     }
1123
1124     /* Increment number of outer iterations */
1125     outeriter        += nri;
1126
1127     /* Update outer/inner flops */
1128     /* ## NB: This is not important, it just affects the flopcount. However, since our preprocessor is */
1129     /* ## primitive and replaces aggressively even in strings inside these directives, we need to      */
1130     /* ## assemble the main part of the name (containing KERNEL/ELEC/VDW) directly in the source.      */
1131     /* #if GEOMETRY_I == 'Water3'            */
1132     /*     #define ISUFFIX '_W3'             */
1133     /* #elif GEOMETRY_I == 'Water4'          */
1134     /*     #define ISUFFIX '_W4'             */
1135     /* #else                                 */
1136     /*     #define ISUFFIX ''                */
1137     /* #endif                                */
1138     /* #if GEOMETRY_J == 'Water3'            */
1139     /*     #define JSUFFIX 'W3'              */
1140     /* #elif GEOMETRY_J == 'Water4'          */
1141     /*     #define JSUFFIX 'W4'              */
1142     /* #else                                 */
1143     /*     #define JSUFFIX ''                */
1144     /* #endif                                */
1145     /* #if 'PotentialAndForce' in KERNEL_VF  */
1146     /*     #define VFSUFFIX  '_VF'           */
1147     /* #elif 'Potential' in KERNEL_VF        */
1148     /*     #define VFSUFFIX '_V'             */
1149     /* #else                                 */
1150     /*     #define VFSUFFIX '_F'             */
1151     /* #endif                                */
1152
1153     /* #if KERNEL_ELEC != 'None' and KERNEL_VDW != 'None' */
1154     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1155     /* #elif KERNEL_ELEC != 'None' */
1156     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1157     /* #else */
1158     inc_nrnb(nrnb,eNR_NBKERNEL_VDW{ISUFFIX}{JSUFFIX}{VFSUFFIX},outeriter*{OUTERFLOPS} + inneriter*{INNERFLOPS});
1159     /* #endif  */
1160 }