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