src/gmxlib/nonbonded/nb_kernel_avx_256_single/nb_kernel_ElecCSTab_VdwLJ_GeomW4P1_avx_256_single.c

   1 /*
   2  * Note: this file was generated by the Gromacs avx_256_single kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 #include "gmx_math_x86_avx_256_single.h"
  34 #include "kernelutil_x86_avx_256_single.h"
  35
  36 /*
  37  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwLJ_GeomW4P1_VF_avx_256_single
  38  * Electrostatics interaction: CubicSplineTable
  39  * VdW interaction:            LennardJones
  40  * Geometry:                   Water4-Particle
  41  * Calculate force/pot:        PotentialAndForce
  42  */
  43 void
  44 nb_kernel_ElecCSTab_VdwLJ_GeomW4P1_VF_avx_256_single
  45                     (t_nblist * gmx_restrict                nlist,
  46                      rvec * gmx_restrict                    xx,
  47                      rvec * gmx_restrict                    ff,
  48                      t_forcerec * gmx_restrict              fr,
  49                      t_mdatoms * gmx_restrict               mdatoms,
  50                      nb_kernel_data_t * gmx_restrict        kernel_data,
  51                      t_nrnb * gmx_restrict                  nrnb)
  52 {
  53     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  54      * just 0 for non-waters.
  55      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
  56      * jnr indices corresponding to data put in the four positions in the SIMD register.
  57      */
  58     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  59     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  60     int              jnrA,jnrB,jnrC,jnrD;
  61     int              jnrE,jnrF,jnrG,jnrH;
  62     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  63     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
  64     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  65     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
  66     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  67     real             rcutoff_scalar;
  68     real             *shiftvec,*fshift,*x,*f;
  69     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
  70     real             scratch[4*DIM];
  71     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  72     real *           vdwioffsetptr0;
  73     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  74     real *           vdwioffsetptr1;
  75     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  76     real *           vdwioffsetptr2;
  77     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  78     real *           vdwioffsetptr3;
  79     __m256           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  80     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
  81     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  82     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  83     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  84     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  85     __m256           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
  86     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
  87     real             *charge;
  88     int              nvdwtype;
  89     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  90     int              *vdwtype;
  91     real             *vdwparam;
  92     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
  93     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
  94     __m256i          vfitab;
  95     __m128i          vfitab_lo,vfitab_hi;
  96     __m128i          ifour       = _mm_set1_epi32(4);
  97     __m256           rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
  98     real             *vftab;
  99     __m256           dummy_mask,cutoff_mask;
 100     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 101     __m256           one     = _mm256_set1_ps(1.0);
 102     __m256           two     = _mm256_set1_ps(2.0);
 103     x                = xx[0];
 104     f                = ff[0];
 105
 106     nri              = nlist->nri;
 107     iinr             = nlist->iinr;
 108     jindex           = nlist->jindex;
 109     jjnr             = nlist->jjnr;
 110     shiftidx         = nlist->shift;
 111     gid              = nlist->gid;
 112     shiftvec         = fr->shift_vec[0];
 113     fshift           = fr->fshift[0];
 114     facel            = _mm256_set1_ps(fr->epsfac);
 115     charge           = mdatoms->chargeA;
 116     nvdwtype         = fr->ntype;
 117     vdwparam         = fr->nbfp;
 118     vdwtype          = mdatoms->typeA;
 119
 120     vftab            = kernel_data->table_elec->data;
 121     vftabscale       = _mm256_set1_ps(kernel_data->table_elec->scale);
 122
 123     /* Setup water-specific parameters */
 124     inr              = nlist->iinr[0];
 125     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 126     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 127     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
 128     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 129
 130     /* Avoid stupid compiler warnings */
 131     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 132     j_coord_offsetA = 0;
 133     j_coord_offsetB = 0;
 134     j_coord_offsetC = 0;
 135     j_coord_offsetD = 0;
 136     j_coord_offsetE = 0;
 137     j_coord_offsetF = 0;
 138     j_coord_offsetG = 0;
 139     j_coord_offsetH = 0;
 140
 141     outeriter        = 0;
 142     inneriter        = 0;
 143
 144     for(iidx=0;iidx<4*DIM;iidx++)
 145     {
 146         scratch[iidx] = 0.0;
 147     }
 148
 149     /* Start outer loop over neighborlists */
 150     for(iidx=0; iidx<nri; iidx++)
 151     {
 152         /* Load shift vector for this list */
 153         i_shift_offset   = DIM*shiftidx[iidx];
 154
 155         /* Load limits for loop over neighbors */
 156         j_index_start    = jindex[iidx];
 157         j_index_end      = jindex[iidx+1];
 158
 159         /* Get outer coordinate index */
 160         inr              = iinr[iidx];
 161         i_coord_offset   = DIM*inr;
 162
 163         /* Load i particle coords and add shift vector */
 164         gmx_mm256_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 165                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 166
 167         fix0             = _mm256_setzero_ps();
 168         fiy0             = _mm256_setzero_ps();
 169         fiz0             = _mm256_setzero_ps();
 170         fix1             = _mm256_setzero_ps();
 171         fiy1             = _mm256_setzero_ps();
 172         fiz1             = _mm256_setzero_ps();
 173         fix2             = _mm256_setzero_ps();
 174         fiy2             = _mm256_setzero_ps();
 175         fiz2             = _mm256_setzero_ps();
 176         fix3             = _mm256_setzero_ps();
 177         fiy3             = _mm256_setzero_ps();
 178         fiz3             = _mm256_setzero_ps();
 179
 180         /* Reset potential sums */
 181         velecsum         = _mm256_setzero_ps();
 182         vvdwsum          = _mm256_setzero_ps();
 183
 184         /* Start inner kernel loop */
 185         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 186         {
 187
 188             /* Get j neighbor index, and coordinate index */
 189             jnrA             = jjnr[jidx];
 190             jnrB             = jjnr[jidx+1];
 191             jnrC             = jjnr[jidx+2];
 192             jnrD             = jjnr[jidx+3];
 193             jnrE             = jjnr[jidx+4];
 194             jnrF             = jjnr[jidx+5];
 195             jnrG             = jjnr[jidx+6];
 196             jnrH             = jjnr[jidx+7];
 197             j_coord_offsetA  = DIM*jnrA;
 198             j_coord_offsetB  = DIM*jnrB;
 199             j_coord_offsetC  = DIM*jnrC;
 200             j_coord_offsetD  = DIM*jnrD;
 201             j_coord_offsetE  = DIM*jnrE;
 202             j_coord_offsetF  = DIM*jnrF;
 203             j_coord_offsetG  = DIM*jnrG;
 204             j_coord_offsetH  = DIM*jnrH;
 205
 206             /* load j atom coordinates */
 207             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 208                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 209                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 210                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 211                                                  &jx0,&jy0,&jz0);
 212
 213             /* Calculate displacement vector */
 214             dx00             = _mm256_sub_ps(ix0,jx0);
 215             dy00             = _mm256_sub_ps(iy0,jy0);
 216             dz00             = _mm256_sub_ps(iz0,jz0);
 217             dx10             = _mm256_sub_ps(ix1,jx0);
 218             dy10             = _mm256_sub_ps(iy1,jy0);
 219             dz10             = _mm256_sub_ps(iz1,jz0);
 220             dx20             = _mm256_sub_ps(ix2,jx0);
 221             dy20             = _mm256_sub_ps(iy2,jy0);
 222             dz20             = _mm256_sub_ps(iz2,jz0);
 223             dx30             = _mm256_sub_ps(ix3,jx0);
 224             dy30             = _mm256_sub_ps(iy3,jy0);
 225             dz30             = _mm256_sub_ps(iz3,jz0);
 226
 227             /* Calculate squared distance and things based on it */
 228             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 229             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 230             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 231             rsq30            = gmx_mm256_calc_rsq_ps(dx30,dy30,dz30);
 232
 233             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 234             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 235             rinv30           = gmx_mm256_invsqrt_ps(rsq30);
 236
 237             rinvsq00         = gmx_mm256_inv_ps(rsq00);
 238
 239             /* Load parameters for j particles */
 240             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 241                                                                  charge+jnrC+0,charge+jnrD+0,
 242                                                                  charge+jnrE+0,charge+jnrF+0,
 243                                                                  charge+jnrG+0,charge+jnrH+0);
 244             vdwjidx0A        = 2*vdwtype[jnrA+0];
 245             vdwjidx0B        = 2*vdwtype[jnrB+0];
 246             vdwjidx0C        = 2*vdwtype[jnrC+0];
 247             vdwjidx0D        = 2*vdwtype[jnrD+0];
 248             vdwjidx0E        = 2*vdwtype[jnrE+0];
 249             vdwjidx0F        = 2*vdwtype[jnrF+0];
 250             vdwjidx0G        = 2*vdwtype[jnrG+0];
 251             vdwjidx0H        = 2*vdwtype[jnrH+0];
 252
 253             fjx0             = _mm256_setzero_ps();
 254             fjy0             = _mm256_setzero_ps();
 255             fjz0             = _mm256_setzero_ps();
 256
 257             /**************************
 258              * CALCULATE INTERACTIONS *
 259              **************************/
 260
 261             /* Compute parameters for interactions between i and j atoms */
 262             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 263                                             vdwioffsetptr0+vdwjidx0B,
 264                                             vdwioffsetptr0+vdwjidx0C,
 265                                             vdwioffsetptr0+vdwjidx0D,
 266                                             vdwioffsetptr0+vdwjidx0E,
 267                                             vdwioffsetptr0+vdwjidx0F,
 268                                             vdwioffsetptr0+vdwjidx0G,
 269                                             vdwioffsetptr0+vdwjidx0H,
 270                                             &c6_00,&c12_00);
 271
 272             /* LENNARD-JONES DISPERSION/REPULSION */
 273
 274             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 275             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
 276             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 277             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
 278             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
 279
 280             /* Update potential sum for this i atom from the interaction with this j atom. */
 281             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 282
 283             fscal            = fvdw;
 284
 285             /* Calculate temporary vectorial force */
 286             tx               = _mm256_mul_ps(fscal,dx00);
 287             ty               = _mm256_mul_ps(fscal,dy00);
 288             tz               = _mm256_mul_ps(fscal,dz00);
 289
 290             /* Update vectorial force */
 291             fix0             = _mm256_add_ps(fix0,tx);
 292             fiy0             = _mm256_add_ps(fiy0,ty);
 293             fiz0             = _mm256_add_ps(fiz0,tz);
 294
 295             fjx0             = _mm256_add_ps(fjx0,tx);
 296             fjy0             = _mm256_add_ps(fjy0,ty);
 297             fjz0             = _mm256_add_ps(fjz0,tz);
 298
 299             /**************************
 300              * CALCULATE INTERACTIONS *
 301              **************************/
 302
 303             r10              = _mm256_mul_ps(rsq10,rinv10);
 304
 305             /* Compute parameters for interactions between i and j atoms */
 306             qq10             = _mm256_mul_ps(iq1,jq0);
 307
 308             /* Calculate table index by multiplying r with table scale and truncate to integer */
 309             rt               = _mm256_mul_ps(r10,vftabscale);
 310             vfitab           = _mm256_cvttps_epi32(rt);
 311             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 312             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 313             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 314             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 315             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 316             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 317
 318             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 319             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 320                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 321             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 322                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 323             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 324                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 325             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 326                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 327             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 328             Heps             = _mm256_mul_ps(vfeps,H);
 329             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 330             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 331             velec            = _mm256_mul_ps(qq10,VV);
 332             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 333             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq10,FF),_mm256_mul_ps(vftabscale,rinv10)));
 334
 335             /* Update potential sum for this i atom from the interaction with this j atom. */
 336             velecsum         = _mm256_add_ps(velecsum,velec);
 337
 338             fscal            = felec;
 339
 340             /* Calculate temporary vectorial force */
 341             tx               = _mm256_mul_ps(fscal,dx10);
 342             ty               = _mm256_mul_ps(fscal,dy10);
 343             tz               = _mm256_mul_ps(fscal,dz10);
 344
 345             /* Update vectorial force */
 346             fix1             = _mm256_add_ps(fix1,tx);
 347             fiy1             = _mm256_add_ps(fiy1,ty);
 348             fiz1             = _mm256_add_ps(fiz1,tz);
 349
 350             fjx0             = _mm256_add_ps(fjx0,tx);
 351             fjy0             = _mm256_add_ps(fjy0,ty);
 352             fjz0             = _mm256_add_ps(fjz0,tz);
 353
 354             /**************************
 355              * CALCULATE INTERACTIONS *
 356              **************************/
 357
 358             r20              = _mm256_mul_ps(rsq20,rinv20);
 359
 360             /* Compute parameters for interactions between i and j atoms */
 361             qq20             = _mm256_mul_ps(iq2,jq0);
 362
 363             /* Calculate table index by multiplying r with table scale and truncate to integer */
 364             rt               = _mm256_mul_ps(r20,vftabscale);
 365             vfitab           = _mm256_cvttps_epi32(rt);
 366             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 367             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 368             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 369             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 370             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 371             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 372
 373             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 374             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 375                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 376             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 377                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 378             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 379                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 380             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 381                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 382             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 383             Heps             = _mm256_mul_ps(vfeps,H);
 384             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 385             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 386             velec            = _mm256_mul_ps(qq20,VV);
 387             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 388             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq20,FF),_mm256_mul_ps(vftabscale,rinv20)));
 389
 390             /* Update potential sum for this i atom from the interaction with this j atom. */
 391             velecsum         = _mm256_add_ps(velecsum,velec);
 392
 393             fscal            = felec;
 394
 395             /* Calculate temporary vectorial force */
 396             tx               = _mm256_mul_ps(fscal,dx20);
 397             ty               = _mm256_mul_ps(fscal,dy20);
 398             tz               = _mm256_mul_ps(fscal,dz20);
 399
 400             /* Update vectorial force */
 401             fix2             = _mm256_add_ps(fix2,tx);
 402             fiy2             = _mm256_add_ps(fiy2,ty);
 403             fiz2             = _mm256_add_ps(fiz2,tz);
 404
 405             fjx0             = _mm256_add_ps(fjx0,tx);
 406             fjy0             = _mm256_add_ps(fjy0,ty);
 407             fjz0             = _mm256_add_ps(fjz0,tz);
 408
 409             /**************************
 410              * CALCULATE INTERACTIONS *
 411              **************************/
 412
 413             r30              = _mm256_mul_ps(rsq30,rinv30);
 414
 415             /* Compute parameters for interactions between i and j atoms */
 416             qq30             = _mm256_mul_ps(iq3,jq0);
 417
 418             /* Calculate table index by multiplying r with table scale and truncate to integer */
 419             rt               = _mm256_mul_ps(r30,vftabscale);
 420             vfitab           = _mm256_cvttps_epi32(rt);
 421             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 422             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 423             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 424             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 425             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 426             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 427
 428             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 429             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 430                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 431             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 432                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 433             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 434                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 435             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 436                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 437             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 438             Heps             = _mm256_mul_ps(vfeps,H);
 439             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 440             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 441             velec            = _mm256_mul_ps(qq30,VV);
 442             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 443             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq30,FF),_mm256_mul_ps(vftabscale,rinv30)));
 444
 445             /* Update potential sum for this i atom from the interaction with this j atom. */
 446             velecsum         = _mm256_add_ps(velecsum,velec);
 447
 448             fscal            = felec;
 449
 450             /* Calculate temporary vectorial force */
 451             tx               = _mm256_mul_ps(fscal,dx30);
 452             ty               = _mm256_mul_ps(fscal,dy30);
 453             tz               = _mm256_mul_ps(fscal,dz30);
 454
 455             /* Update vectorial force */
 456             fix3             = _mm256_add_ps(fix3,tx);
 457             fiy3             = _mm256_add_ps(fiy3,ty);
 458             fiz3             = _mm256_add_ps(fiz3,tz);
 459
 460             fjx0             = _mm256_add_ps(fjx0,tx);
 461             fjy0             = _mm256_add_ps(fjy0,ty);
 462             fjz0             = _mm256_add_ps(fjz0,tz);
 463
 464             fjptrA             = f+j_coord_offsetA;
 465             fjptrB             = f+j_coord_offsetB;
 466             fjptrC             = f+j_coord_offsetC;
 467             fjptrD             = f+j_coord_offsetD;
 468             fjptrE             = f+j_coord_offsetE;
 469             fjptrF             = f+j_coord_offsetF;
 470             fjptrG             = f+j_coord_offsetG;
 471             fjptrH             = f+j_coord_offsetH;
 472
 473             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
 474
 475             /* Inner loop uses 164 flops */
 476         }
 477
 478         if(jidx<j_index_end)
 479         {
 480
 481             /* Get j neighbor index, and coordinate index */
 482             jnrlistA         = jjnr[jidx];
 483             jnrlistB         = jjnr[jidx+1];
 484             jnrlistC         = jjnr[jidx+2];
 485             jnrlistD         = jjnr[jidx+3];
 486             jnrlistE         = jjnr[jidx+4];
 487             jnrlistF         = jjnr[jidx+5];
 488             jnrlistG         = jjnr[jidx+6];
 489             jnrlistH         = jjnr[jidx+7];
 490             /* Sign of each element will be negative for non-real atoms.
 491              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 492              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 493              */
 494             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
 495                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
 496
 497             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 498             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 499             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 500             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 501             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
 502             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
 503             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
 504             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
 505             j_coord_offsetA  = DIM*jnrA;
 506             j_coord_offsetB  = DIM*jnrB;
 507             j_coord_offsetC  = DIM*jnrC;
 508             j_coord_offsetD  = DIM*jnrD;
 509             j_coord_offsetE  = DIM*jnrE;
 510             j_coord_offsetF  = DIM*jnrF;
 511             j_coord_offsetG  = DIM*jnrG;
 512             j_coord_offsetH  = DIM*jnrH;
 513
 514             /* load j atom coordinates */
 515             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 516                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 517                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 518                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 519                                                  &jx0,&jy0,&jz0);
 520
 521             /* Calculate displacement vector */
 522             dx00             = _mm256_sub_ps(ix0,jx0);
 523             dy00             = _mm256_sub_ps(iy0,jy0);
 524             dz00             = _mm256_sub_ps(iz0,jz0);
 525             dx10             = _mm256_sub_ps(ix1,jx0);
 526             dy10             = _mm256_sub_ps(iy1,jy0);
 527             dz10             = _mm256_sub_ps(iz1,jz0);
 528             dx20             = _mm256_sub_ps(ix2,jx0);
 529             dy20             = _mm256_sub_ps(iy2,jy0);
 530             dz20             = _mm256_sub_ps(iz2,jz0);
 531             dx30             = _mm256_sub_ps(ix3,jx0);
 532             dy30             = _mm256_sub_ps(iy3,jy0);
 533             dz30             = _mm256_sub_ps(iz3,jz0);
 534
 535             /* Calculate squared distance and things based on it */
 536             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
 537             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
 538             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
 539             rsq30            = gmx_mm256_calc_rsq_ps(dx30,dy30,dz30);
 540
 541             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
 542             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
 543             rinv30           = gmx_mm256_invsqrt_ps(rsq30);
 544
 545             rinvsq00         = gmx_mm256_inv_ps(rsq00);
 546
 547             /* Load parameters for j particles */
 548             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 549                                                                  charge+jnrC+0,charge+jnrD+0,
 550                                                                  charge+jnrE+0,charge+jnrF+0,
 551                                                                  charge+jnrG+0,charge+jnrH+0);
 552             vdwjidx0A        = 2*vdwtype[jnrA+0];
 553             vdwjidx0B        = 2*vdwtype[jnrB+0];
 554             vdwjidx0C        = 2*vdwtype[jnrC+0];
 555             vdwjidx0D        = 2*vdwtype[jnrD+0];
 556             vdwjidx0E        = 2*vdwtype[jnrE+0];
 557             vdwjidx0F        = 2*vdwtype[jnrF+0];
 558             vdwjidx0G        = 2*vdwtype[jnrG+0];
 559             vdwjidx0H        = 2*vdwtype[jnrH+0];
 560
 561             fjx0             = _mm256_setzero_ps();
 562             fjy0             = _mm256_setzero_ps();
 563             fjz0             = _mm256_setzero_ps();
 564
 565             /**************************
 566              * CALCULATE INTERACTIONS *
 567              **************************/
 568
 569             /* Compute parameters for interactions between i and j atoms */
 570             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
 571                                             vdwioffsetptr0+vdwjidx0B,
 572                                             vdwioffsetptr0+vdwjidx0C,
 573                                             vdwioffsetptr0+vdwjidx0D,
 574                                             vdwioffsetptr0+vdwjidx0E,
 575                                             vdwioffsetptr0+vdwjidx0F,
 576                                             vdwioffsetptr0+vdwjidx0G,
 577                                             vdwioffsetptr0+vdwjidx0H,
 578                                             &c6_00,&c12_00);
 579
 580             /* LENNARD-JONES DISPERSION/REPULSION */
 581
 582             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 583             vvdw6            = _mm256_mul_ps(c6_00,rinvsix);
 584             vvdw12           = _mm256_mul_ps(c12_00,_mm256_mul_ps(rinvsix,rinvsix));
 585             vvdw             = _mm256_sub_ps( _mm256_mul_ps(vvdw12,one_twelfth) , _mm256_mul_ps(vvdw6,one_sixth) );
 586             fvdw             = _mm256_mul_ps(_mm256_sub_ps(vvdw12,vvdw6),rinvsq00);
 587
 588             /* Update potential sum for this i atom from the interaction with this j atom. */
 589             vvdw             = _mm256_andnot_ps(dummy_mask,vvdw);
 590             vvdwsum          = _mm256_add_ps(vvdwsum,vvdw);
 591
 592             fscal            = fvdw;
 593
 594             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 595
 596             /* Calculate temporary vectorial force */
 597             tx               = _mm256_mul_ps(fscal,dx00);
 598             ty               = _mm256_mul_ps(fscal,dy00);
 599             tz               = _mm256_mul_ps(fscal,dz00);
 600
 601             /* Update vectorial force */
 602             fix0             = _mm256_add_ps(fix0,tx);
 603             fiy0             = _mm256_add_ps(fiy0,ty);
 604             fiz0             = _mm256_add_ps(fiz0,tz);
 605
 606             fjx0             = _mm256_add_ps(fjx0,tx);
 607             fjy0             = _mm256_add_ps(fjy0,ty);
 608             fjz0             = _mm256_add_ps(fjz0,tz);
 609
 610             /**************************
 611              * CALCULATE INTERACTIONS *
 612              **************************/
 613
 614             r10              = _mm256_mul_ps(rsq10,rinv10);
 615             r10              = _mm256_andnot_ps(dummy_mask,r10);
 616
 617             /* Compute parameters for interactions between i and j atoms */
 618             qq10             = _mm256_mul_ps(iq1,jq0);
 619
 620             /* Calculate table index by multiplying r with table scale and truncate to integer */
 621             rt               = _mm256_mul_ps(r10,vftabscale);
 622             vfitab           = _mm256_cvttps_epi32(rt);
 623             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 624             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 625             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 626             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 627             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 628             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 629
 630             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 631             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 632                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 633             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 634                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 635             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 636                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 637             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 638                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 639             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 640             Heps             = _mm256_mul_ps(vfeps,H);
 641             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 642             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 643             velec            = _mm256_mul_ps(qq10,VV);
 644             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 645             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq10,FF),_mm256_mul_ps(vftabscale,rinv10)));
 646
 647             /* Update potential sum for this i atom from the interaction with this j atom. */
 648             velec            = _mm256_andnot_ps(dummy_mask,velec);
 649             velecsum         = _mm256_add_ps(velecsum,velec);
 650
 651             fscal            = felec;
 652
 653             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 654
 655             /* Calculate temporary vectorial force */
 656             tx               = _mm256_mul_ps(fscal,dx10);
 657             ty               = _mm256_mul_ps(fscal,dy10);
 658             tz               = _mm256_mul_ps(fscal,dz10);
 659
 660             /* Update vectorial force */
 661             fix1             = _mm256_add_ps(fix1,tx);
 662             fiy1             = _mm256_add_ps(fiy1,ty);
 663             fiz1             = _mm256_add_ps(fiz1,tz);
 664
 665             fjx0             = _mm256_add_ps(fjx0,tx);
 666             fjy0             = _mm256_add_ps(fjy0,ty);
 667             fjz0             = _mm256_add_ps(fjz0,tz);
 668
 669             /**************************
 670              * CALCULATE INTERACTIONS *
 671              **************************/
 672
 673             r20              = _mm256_mul_ps(rsq20,rinv20);
 674             r20              = _mm256_andnot_ps(dummy_mask,r20);
 675
 676             /* Compute parameters for interactions between i and j atoms */
 677             qq20             = _mm256_mul_ps(iq2,jq0);
 678
 679             /* Calculate table index by multiplying r with table scale and truncate to integer */
 680             rt               = _mm256_mul_ps(r20,vftabscale);
 681             vfitab           = _mm256_cvttps_epi32(rt);
 682             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 683             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 684             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 685             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 686             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 687             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 688
 689             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 690             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 691                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 692             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 693                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 694             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 695                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 696             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 697                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 698             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 699             Heps             = _mm256_mul_ps(vfeps,H);
 700             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 701             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 702             velec            = _mm256_mul_ps(qq20,VV);
 703             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 704             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq20,FF),_mm256_mul_ps(vftabscale,rinv20)));
 705
 706             /* Update potential sum for this i atom from the interaction with this j atom. */
 707             velec            = _mm256_andnot_ps(dummy_mask,velec);
 708             velecsum         = _mm256_add_ps(velecsum,velec);
 709
 710             fscal            = felec;
 711
 712             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 713
 714             /* Calculate temporary vectorial force */
 715             tx               = _mm256_mul_ps(fscal,dx20);
 716             ty               = _mm256_mul_ps(fscal,dy20);
 717             tz               = _mm256_mul_ps(fscal,dz20);
 718
 719             /* Update vectorial force */
 720             fix2             = _mm256_add_ps(fix2,tx);
 721             fiy2             = _mm256_add_ps(fiy2,ty);
 722             fiz2             = _mm256_add_ps(fiz2,tz);
 723
 724             fjx0             = _mm256_add_ps(fjx0,tx);
 725             fjy0             = _mm256_add_ps(fjy0,ty);
 726             fjz0             = _mm256_add_ps(fjz0,tz);
 727
 728             /**************************
 729              * CALCULATE INTERACTIONS *
 730              **************************/
 731
 732             r30              = _mm256_mul_ps(rsq30,rinv30);
 733             r30              = _mm256_andnot_ps(dummy_mask,r30);
 734
 735             /* Compute parameters for interactions between i and j atoms */
 736             qq30             = _mm256_mul_ps(iq3,jq0);
 737
 738             /* Calculate table index by multiplying r with table scale and truncate to integer */
 739             rt               = _mm256_mul_ps(r30,vftabscale);
 740             vfitab           = _mm256_cvttps_epi32(rt);
 741             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
 742             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
 743             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
 744             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
 745             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
 746             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
 747
 748             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 749             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
 750                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
 751             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
 752                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
 753             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
 754                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
 755             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
 756                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
 757             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
 758             Heps             = _mm256_mul_ps(vfeps,H);
 759             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
 760             VV               = _mm256_add_ps(Y,_mm256_mul_ps(vfeps,Fp));
 761             velec            = _mm256_mul_ps(qq30,VV);
 762             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
 763             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq30,FF),_mm256_mul_ps(vftabscale,rinv30)));
 764
 765             /* Update potential sum for this i atom from the interaction with this j atom. */
 766             velec            = _mm256_andnot_ps(dummy_mask,velec);
 767             velecsum         = _mm256_add_ps(velecsum,velec);
 768
 769             fscal            = felec;
 770
 771             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
 772
 773             /* Calculate temporary vectorial force */
 774             tx               = _mm256_mul_ps(fscal,dx30);
 775             ty               = _mm256_mul_ps(fscal,dy30);
 776             tz               = _mm256_mul_ps(fscal,dz30);
 777
 778             /* Update vectorial force */
 779             fix3             = _mm256_add_ps(fix3,tx);
 780             fiy3             = _mm256_add_ps(fiy3,ty);
 781             fiz3             = _mm256_add_ps(fiz3,tz);
 782
 783             fjx0             = _mm256_add_ps(fjx0,tx);
 784             fjy0             = _mm256_add_ps(fjy0,ty);
 785             fjz0             = _mm256_add_ps(fjz0,tz);
 786
 787             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 788             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 789             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 790             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 791             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
 792             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
 793             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
 794             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
 795
 796             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
 797
 798             /* Inner loop uses 167 flops */
 799         }
 800
 801         /* End of innermost loop */
 802
 803         gmx_mm256_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 804                                                  f+i_coord_offset,fshift+i_shift_offset);
 805
 806         ggid                        = gid[iidx];
 807         /* Update potential energies */
 808         gmx_mm256_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 809         gmx_mm256_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 810
 811         /* Increment number of inner iterations */
 812         inneriter                  += j_index_end - j_index_start;
 813
 814         /* Outer loop uses 26 flops */
 815     }
 816
 817     /* Increment number of outer iterations */
 818     outeriter        += nri;
 819
 820     /* Update outer/inner flops */
 821
 822     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*167);
 823 }
 824 /*
 825  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwLJ_GeomW4P1_F_avx_256_single
 826  * Electrostatics interaction: CubicSplineTable
 827  * VdW interaction:            LennardJones
 828  * Geometry:                   Water4-Particle
 829  * Calculate force/pot:        Force
 830  */
 831 void
 832 nb_kernel_ElecCSTab_VdwLJ_GeomW4P1_F_avx_256_single
 833                     (t_nblist * gmx_restrict                nlist,
 834                      rvec * gmx_restrict                    xx,
 835                      rvec * gmx_restrict                    ff,
 836                      t_forcerec * gmx_restrict              fr,
 837                      t_mdatoms * gmx_restrict               mdatoms,
 838                      nb_kernel_data_t * gmx_restrict        kernel_data,
 839                      t_nrnb * gmx_restrict                  nrnb)
 840 {
 841     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 842      * just 0 for non-waters.
 843      * Suffixes A,B,C,D,E,F,G,H refer to j loop unrolling done with AVX, e.g. for the eight different
 844      * jnr indices corresponding to data put in the four positions in the SIMD register.
 845      */
 846     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 847     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 848     int              jnrA,jnrB,jnrC,jnrD;
 849     int              jnrE,jnrF,jnrG,jnrH;
 850     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 851     int              jnrlistE,jnrlistF,jnrlistG,jnrlistH;
 852     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 853     int              j_coord_offsetE,j_coord_offsetF,j_coord_offsetG,j_coord_offsetH;
 854     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 855     real             rcutoff_scalar;
 856     real             *shiftvec,*fshift,*x,*f;
 857     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD,*fjptrE,*fjptrF,*fjptrG,*fjptrH;
 858     real             scratch[4*DIM];
 859     __m256           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 860     real *           vdwioffsetptr0;
 861     __m256           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 862     real *           vdwioffsetptr1;
 863     __m256           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 864     real *           vdwioffsetptr2;
 865     __m256           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 866     real *           vdwioffsetptr3;
 867     __m256           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 868     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D,vdwjidx0E,vdwjidx0F,vdwjidx0G,vdwjidx0H;
 869     __m256           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 870     __m256           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 871     __m256           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 872     __m256           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 873     __m256           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 874     __m256           velec,felec,velecsum,facel,crf,krf,krf2;
 875     real             *charge;
 876     int              nvdwtype;
 877     __m256           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 878     int              *vdwtype;
 879     real             *vdwparam;
 880     __m256           one_sixth   = _mm256_set1_ps(1.0/6.0);
 881     __m256           one_twelfth = _mm256_set1_ps(1.0/12.0);
 882     __m256i          vfitab;
 883     __m128i          vfitab_lo,vfitab_hi;
 884     __m128i          ifour       = _mm_set1_epi32(4);
 885     __m256           rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF;
 886     real             *vftab;
 887     __m256           dummy_mask,cutoff_mask;
 888     __m256           signbit = _mm256_castsi256_ps( _mm256_set1_epi32(0x80000000) );
 889     __m256           one     = _mm256_set1_ps(1.0);
 890     __m256           two     = _mm256_set1_ps(2.0);
 891     x                = xx[0];
 892     f                = ff[0];
 893
 894     nri              = nlist->nri;
 895     iinr             = nlist->iinr;
 896     jindex           = nlist->jindex;
 897     jjnr             = nlist->jjnr;
 898     shiftidx         = nlist->shift;
 899     gid              = nlist->gid;
 900     shiftvec         = fr->shift_vec[0];
 901     fshift           = fr->fshift[0];
 902     facel            = _mm256_set1_ps(fr->epsfac);
 903     charge           = mdatoms->chargeA;
 904     nvdwtype         = fr->ntype;
 905     vdwparam         = fr->nbfp;
 906     vdwtype          = mdatoms->typeA;
 907
 908     vftab            = kernel_data->table_elec->data;
 909     vftabscale       = _mm256_set1_ps(kernel_data->table_elec->scale);
 910
 911     /* Setup water-specific parameters */
 912     inr              = nlist->iinr[0];
 913     iq1              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+1]));
 914     iq2              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+2]));
 915     iq3              = _mm256_mul_ps(facel,_mm256_set1_ps(charge[inr+3]));
 916     vdwioffsetptr0   = vdwparam+2*nvdwtype*vdwtype[inr+0];
 917
 918     /* Avoid stupid compiler warnings */
 919     jnrA = jnrB = jnrC = jnrD = jnrE = jnrF = jnrG = jnrH = 0;
 920     j_coord_offsetA = 0;
 921     j_coord_offsetB = 0;
 922     j_coord_offsetC = 0;
 923     j_coord_offsetD = 0;
 924     j_coord_offsetE = 0;
 925     j_coord_offsetF = 0;
 926     j_coord_offsetG = 0;
 927     j_coord_offsetH = 0;
 928
 929     outeriter        = 0;
 930     inneriter        = 0;
 931
 932     for(iidx=0;iidx<4*DIM;iidx++)
 933     {
 934         scratch[iidx] = 0.0;
 935     }
 936
 937     /* Start outer loop over neighborlists */
 938     for(iidx=0; iidx<nri; iidx++)
 939     {
 940         /* Load shift vector for this list */
 941         i_shift_offset   = DIM*shiftidx[iidx];
 942
 943         /* Load limits for loop over neighbors */
 944         j_index_start    = jindex[iidx];
 945         j_index_end      = jindex[iidx+1];
 946
 947         /* Get outer coordinate index */
 948         inr              = iinr[iidx];
 949         i_coord_offset   = DIM*inr;
 950
 951         /* Load i particle coords and add shift vector */
 952         gmx_mm256_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 953                                                     &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 954
 955         fix0             = _mm256_setzero_ps();
 956         fiy0             = _mm256_setzero_ps();
 957         fiz0             = _mm256_setzero_ps();
 958         fix1             = _mm256_setzero_ps();
 959         fiy1             = _mm256_setzero_ps();
 960         fiz1             = _mm256_setzero_ps();
 961         fix2             = _mm256_setzero_ps();
 962         fiy2             = _mm256_setzero_ps();
 963         fiz2             = _mm256_setzero_ps();
 964         fix3             = _mm256_setzero_ps();
 965         fiy3             = _mm256_setzero_ps();
 966         fiz3             = _mm256_setzero_ps();
 967
 968         /* Start inner kernel loop */
 969         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+7]>=0; jidx+=8)
 970         {
 971
 972             /* Get j neighbor index, and coordinate index */
 973             jnrA             = jjnr[jidx];
 974             jnrB             = jjnr[jidx+1];
 975             jnrC             = jjnr[jidx+2];
 976             jnrD             = jjnr[jidx+3];
 977             jnrE             = jjnr[jidx+4];
 978             jnrF             = jjnr[jidx+5];
 979             jnrG             = jjnr[jidx+6];
 980             jnrH             = jjnr[jidx+7];
 981             j_coord_offsetA  = DIM*jnrA;
 982             j_coord_offsetB  = DIM*jnrB;
 983             j_coord_offsetC  = DIM*jnrC;
 984             j_coord_offsetD  = DIM*jnrD;
 985             j_coord_offsetE  = DIM*jnrE;
 986             j_coord_offsetF  = DIM*jnrF;
 987             j_coord_offsetG  = DIM*jnrG;
 988             j_coord_offsetH  = DIM*jnrH;
 989
 990             /* load j atom coordinates */
 991             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 992                                                  x+j_coord_offsetC,x+j_coord_offsetD,
 993                                                  x+j_coord_offsetE,x+j_coord_offsetF,
 994                                                  x+j_coord_offsetG,x+j_coord_offsetH,
 995                                                  &jx0,&jy0,&jz0);
 996
 997             /* Calculate displacement vector */
 998             dx00             = _mm256_sub_ps(ix0,jx0);
 999             dy00             = _mm256_sub_ps(iy0,jy0);
1000             dz00             = _mm256_sub_ps(iz0,jz0);
1001             dx10             = _mm256_sub_ps(ix1,jx0);
1002             dy10             = _mm256_sub_ps(iy1,jy0);
1003             dz10             = _mm256_sub_ps(iz1,jz0);
1004             dx20             = _mm256_sub_ps(ix2,jx0);
1005             dy20             = _mm256_sub_ps(iy2,jy0);
1006             dz20             = _mm256_sub_ps(iz2,jz0);
1007             dx30             = _mm256_sub_ps(ix3,jx0);
1008             dy30             = _mm256_sub_ps(iy3,jy0);
1009             dz30             = _mm256_sub_ps(iz3,jz0);
1010
1011             /* Calculate squared distance and things based on it */
1012             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1013             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1014             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1015             rsq30            = gmx_mm256_calc_rsq_ps(dx30,dy30,dz30);
1016
1017             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1018             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1019             rinv30           = gmx_mm256_invsqrt_ps(rsq30);
1020
1021             rinvsq00         = gmx_mm256_inv_ps(rsq00);
1022
1023             /* Load parameters for j particles */
1024             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1025                                                                  charge+jnrC+0,charge+jnrD+0,
1026                                                                  charge+jnrE+0,charge+jnrF+0,
1027                                                                  charge+jnrG+0,charge+jnrH+0);
1028             vdwjidx0A        = 2*vdwtype[jnrA+0];
1029             vdwjidx0B        = 2*vdwtype[jnrB+0];
1030             vdwjidx0C        = 2*vdwtype[jnrC+0];
1031             vdwjidx0D        = 2*vdwtype[jnrD+0];
1032             vdwjidx0E        = 2*vdwtype[jnrE+0];
1033             vdwjidx0F        = 2*vdwtype[jnrF+0];
1034             vdwjidx0G        = 2*vdwtype[jnrG+0];
1035             vdwjidx0H        = 2*vdwtype[jnrH+0];
1036
1037             fjx0             = _mm256_setzero_ps();
1038             fjy0             = _mm256_setzero_ps();
1039             fjz0             = _mm256_setzero_ps();
1040
1041             /**************************
1042              * CALCULATE INTERACTIONS *
1043              **************************/
1044
1045             /* Compute parameters for interactions between i and j atoms */
1046             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
1047                                             vdwioffsetptr0+vdwjidx0B,
1048                                             vdwioffsetptr0+vdwjidx0C,
1049                                             vdwioffsetptr0+vdwjidx0D,
1050                                             vdwioffsetptr0+vdwjidx0E,
1051                                             vdwioffsetptr0+vdwjidx0F,
1052                                             vdwioffsetptr0+vdwjidx0G,
1053                                             vdwioffsetptr0+vdwjidx0H,
1054                                             &c6_00,&c12_00);
1055
1056             /* LENNARD-JONES DISPERSION/REPULSION */
1057
1058             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1059             fvdw             = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
1060
1061             fscal            = fvdw;
1062
1063             /* Calculate temporary vectorial force */
1064             tx               = _mm256_mul_ps(fscal,dx00);
1065             ty               = _mm256_mul_ps(fscal,dy00);
1066             tz               = _mm256_mul_ps(fscal,dz00);
1067
1068             /* Update vectorial force */
1069             fix0             = _mm256_add_ps(fix0,tx);
1070             fiy0             = _mm256_add_ps(fiy0,ty);
1071             fiz0             = _mm256_add_ps(fiz0,tz);
1072
1073             fjx0             = _mm256_add_ps(fjx0,tx);
1074             fjy0             = _mm256_add_ps(fjy0,ty);
1075             fjz0             = _mm256_add_ps(fjz0,tz);
1076
1077             /**************************
1078              * CALCULATE INTERACTIONS *
1079              **************************/
1080
1081             r10              = _mm256_mul_ps(rsq10,rinv10);
1082
1083             /* Compute parameters for interactions between i and j atoms */
1084             qq10             = _mm256_mul_ps(iq1,jq0);
1085
1086             /* Calculate table index by multiplying r with table scale and truncate to integer */
1087             rt               = _mm256_mul_ps(r10,vftabscale);
1088             vfitab           = _mm256_cvttps_epi32(rt);
1089             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1090             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1091             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1092             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1093             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1094             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1095
1096             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1097             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1098                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1099             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1100                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1101             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1102                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1103             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1104                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1105             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1106             Heps             = _mm256_mul_ps(vfeps,H);
1107             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1108             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1109             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq10,FF),_mm256_mul_ps(vftabscale,rinv10)));
1110
1111             fscal            = felec;
1112
1113             /* Calculate temporary vectorial force */
1114             tx               = _mm256_mul_ps(fscal,dx10);
1115             ty               = _mm256_mul_ps(fscal,dy10);
1116             tz               = _mm256_mul_ps(fscal,dz10);
1117
1118             /* Update vectorial force */
1119             fix1             = _mm256_add_ps(fix1,tx);
1120             fiy1             = _mm256_add_ps(fiy1,ty);
1121             fiz1             = _mm256_add_ps(fiz1,tz);
1122
1123             fjx0             = _mm256_add_ps(fjx0,tx);
1124             fjy0             = _mm256_add_ps(fjy0,ty);
1125             fjz0             = _mm256_add_ps(fjz0,tz);
1126
1127             /**************************
1128              * CALCULATE INTERACTIONS *
1129              **************************/
1130
1131             r20              = _mm256_mul_ps(rsq20,rinv20);
1132
1133             /* Compute parameters for interactions between i and j atoms */
1134             qq20             = _mm256_mul_ps(iq2,jq0);
1135
1136             /* Calculate table index by multiplying r with table scale and truncate to integer */
1137             rt               = _mm256_mul_ps(r20,vftabscale);
1138             vfitab           = _mm256_cvttps_epi32(rt);
1139             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1140             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1141             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1142             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1143             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1144             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1145
1146             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1147             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1148                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1149             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1150                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1151             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1152                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1153             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1154                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1155             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1156             Heps             = _mm256_mul_ps(vfeps,H);
1157             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1158             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1159             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq20,FF),_mm256_mul_ps(vftabscale,rinv20)));
1160
1161             fscal            = felec;
1162
1163             /* Calculate temporary vectorial force */
1164             tx               = _mm256_mul_ps(fscal,dx20);
1165             ty               = _mm256_mul_ps(fscal,dy20);
1166             tz               = _mm256_mul_ps(fscal,dz20);
1167
1168             /* Update vectorial force */
1169             fix2             = _mm256_add_ps(fix2,tx);
1170             fiy2             = _mm256_add_ps(fiy2,ty);
1171             fiz2             = _mm256_add_ps(fiz2,tz);
1172
1173             fjx0             = _mm256_add_ps(fjx0,tx);
1174             fjy0             = _mm256_add_ps(fjy0,ty);
1175             fjz0             = _mm256_add_ps(fjz0,tz);
1176
1177             /**************************
1178              * CALCULATE INTERACTIONS *
1179              **************************/
1180
1181             r30              = _mm256_mul_ps(rsq30,rinv30);
1182
1183             /* Compute parameters for interactions between i and j atoms */
1184             qq30             = _mm256_mul_ps(iq3,jq0);
1185
1186             /* Calculate table index by multiplying r with table scale and truncate to integer */
1187             rt               = _mm256_mul_ps(r30,vftabscale);
1188             vfitab           = _mm256_cvttps_epi32(rt);
1189             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1190             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1191             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1192             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1193             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1194             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1195
1196             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1197             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1198                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1199             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1200                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1201             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1202                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1203             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1204                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1205             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1206             Heps             = _mm256_mul_ps(vfeps,H);
1207             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1208             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1209             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq30,FF),_mm256_mul_ps(vftabscale,rinv30)));
1210
1211             fscal            = felec;
1212
1213             /* Calculate temporary vectorial force */
1214             tx               = _mm256_mul_ps(fscal,dx30);
1215             ty               = _mm256_mul_ps(fscal,dy30);
1216             tz               = _mm256_mul_ps(fscal,dz30);
1217
1218             /* Update vectorial force */
1219             fix3             = _mm256_add_ps(fix3,tx);
1220             fiy3             = _mm256_add_ps(fiy3,ty);
1221             fiz3             = _mm256_add_ps(fiz3,tz);
1222
1223             fjx0             = _mm256_add_ps(fjx0,tx);
1224             fjy0             = _mm256_add_ps(fjy0,ty);
1225             fjz0             = _mm256_add_ps(fjz0,tz);
1226
1227             fjptrA             = f+j_coord_offsetA;
1228             fjptrB             = f+j_coord_offsetB;
1229             fjptrC             = f+j_coord_offsetC;
1230             fjptrD             = f+j_coord_offsetD;
1231             fjptrE             = f+j_coord_offsetE;
1232             fjptrF             = f+j_coord_offsetF;
1233             fjptrG             = f+j_coord_offsetG;
1234             fjptrH             = f+j_coord_offsetH;
1235
1236             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1237
1238             /* Inner loop uses 147 flops */
1239         }
1240
1241         if(jidx<j_index_end)
1242         {
1243
1244             /* Get j neighbor index, and coordinate index */
1245             jnrlistA         = jjnr[jidx];
1246             jnrlistB         = jjnr[jidx+1];
1247             jnrlistC         = jjnr[jidx+2];
1248             jnrlistD         = jjnr[jidx+3];
1249             jnrlistE         = jjnr[jidx+4];
1250             jnrlistF         = jjnr[jidx+5];
1251             jnrlistG         = jjnr[jidx+6];
1252             jnrlistH         = jjnr[jidx+7];
1253             /* Sign of each element will be negative for non-real atoms.
1254              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1255              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1256              */
1257             dummy_mask = gmx_mm256_set_m128(gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx+4)),_mm_setzero_si128())),
1258                                             gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128())));
1259
1260             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1261             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1262             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1263             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1264             jnrE       = (jnrlistE>=0) ? jnrlistE : 0;
1265             jnrF       = (jnrlistF>=0) ? jnrlistF : 0;
1266             jnrG       = (jnrlistG>=0) ? jnrlistG : 0;
1267             jnrH       = (jnrlistH>=0) ? jnrlistH : 0;
1268             j_coord_offsetA  = DIM*jnrA;
1269             j_coord_offsetB  = DIM*jnrB;
1270             j_coord_offsetC  = DIM*jnrC;
1271             j_coord_offsetD  = DIM*jnrD;
1272             j_coord_offsetE  = DIM*jnrE;
1273             j_coord_offsetF  = DIM*jnrF;
1274             j_coord_offsetG  = DIM*jnrG;
1275             j_coord_offsetH  = DIM*jnrH;
1276
1277             /* load j atom coordinates */
1278             gmx_mm256_load_1rvec_8ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1279                                                  x+j_coord_offsetC,x+j_coord_offsetD,
1280                                                  x+j_coord_offsetE,x+j_coord_offsetF,
1281                                                  x+j_coord_offsetG,x+j_coord_offsetH,
1282                                                  &jx0,&jy0,&jz0);
1283
1284             /* Calculate displacement vector */
1285             dx00             = _mm256_sub_ps(ix0,jx0);
1286             dy00             = _mm256_sub_ps(iy0,jy0);
1287             dz00             = _mm256_sub_ps(iz0,jz0);
1288             dx10             = _mm256_sub_ps(ix1,jx0);
1289             dy10             = _mm256_sub_ps(iy1,jy0);
1290             dz10             = _mm256_sub_ps(iz1,jz0);
1291             dx20             = _mm256_sub_ps(ix2,jx0);
1292             dy20             = _mm256_sub_ps(iy2,jy0);
1293             dz20             = _mm256_sub_ps(iz2,jz0);
1294             dx30             = _mm256_sub_ps(ix3,jx0);
1295             dy30             = _mm256_sub_ps(iy3,jy0);
1296             dz30             = _mm256_sub_ps(iz3,jz0);
1297
1298             /* Calculate squared distance and things based on it */
1299             rsq00            = gmx_mm256_calc_rsq_ps(dx00,dy00,dz00);
1300             rsq10            = gmx_mm256_calc_rsq_ps(dx10,dy10,dz10);
1301             rsq20            = gmx_mm256_calc_rsq_ps(dx20,dy20,dz20);
1302             rsq30            = gmx_mm256_calc_rsq_ps(dx30,dy30,dz30);
1303
1304             rinv10           = gmx_mm256_invsqrt_ps(rsq10);
1305             rinv20           = gmx_mm256_invsqrt_ps(rsq20);
1306             rinv30           = gmx_mm256_invsqrt_ps(rsq30);
1307
1308             rinvsq00         = gmx_mm256_inv_ps(rsq00);
1309
1310             /* Load parameters for j particles */
1311             jq0              = gmx_mm256_load_8real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1312                                                                  charge+jnrC+0,charge+jnrD+0,
1313                                                                  charge+jnrE+0,charge+jnrF+0,
1314                                                                  charge+jnrG+0,charge+jnrH+0);
1315             vdwjidx0A        = 2*vdwtype[jnrA+0];
1316             vdwjidx0B        = 2*vdwtype[jnrB+0];
1317             vdwjidx0C        = 2*vdwtype[jnrC+0];
1318             vdwjidx0D        = 2*vdwtype[jnrD+0];
1319             vdwjidx0E        = 2*vdwtype[jnrE+0];
1320             vdwjidx0F        = 2*vdwtype[jnrF+0];
1321             vdwjidx0G        = 2*vdwtype[jnrG+0];
1322             vdwjidx0H        = 2*vdwtype[jnrH+0];
1323
1324             fjx0             = _mm256_setzero_ps();
1325             fjy0             = _mm256_setzero_ps();
1326             fjz0             = _mm256_setzero_ps();
1327
1328             /**************************
1329              * CALCULATE INTERACTIONS *
1330              **************************/
1331
1332             /* Compute parameters for interactions between i and j atoms */
1333             gmx_mm256_load_8pair_swizzle_ps(vdwioffsetptr0+vdwjidx0A,
1334                                             vdwioffsetptr0+vdwjidx0B,
1335                                             vdwioffsetptr0+vdwjidx0C,
1336                                             vdwioffsetptr0+vdwjidx0D,
1337                                             vdwioffsetptr0+vdwjidx0E,
1338                                             vdwioffsetptr0+vdwjidx0F,
1339                                             vdwioffsetptr0+vdwjidx0G,
1340                                             vdwioffsetptr0+vdwjidx0H,
1341                                             &c6_00,&c12_00);
1342
1343             /* LENNARD-JONES DISPERSION/REPULSION */
1344
1345             rinvsix          = _mm256_mul_ps(_mm256_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1346             fvdw             = _mm256_mul_ps(_mm256_sub_ps(_mm256_mul_ps(c12_00,rinvsix),c6_00),_mm256_mul_ps(rinvsix,rinvsq00));
1347
1348             fscal            = fvdw;
1349
1350             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1351
1352             /* Calculate temporary vectorial force */
1353             tx               = _mm256_mul_ps(fscal,dx00);
1354             ty               = _mm256_mul_ps(fscal,dy00);
1355             tz               = _mm256_mul_ps(fscal,dz00);
1356
1357             /* Update vectorial force */
1358             fix0             = _mm256_add_ps(fix0,tx);
1359             fiy0             = _mm256_add_ps(fiy0,ty);
1360             fiz0             = _mm256_add_ps(fiz0,tz);
1361
1362             fjx0             = _mm256_add_ps(fjx0,tx);
1363             fjy0             = _mm256_add_ps(fjy0,ty);
1364             fjz0             = _mm256_add_ps(fjz0,tz);
1365
1366             /**************************
1367              * CALCULATE INTERACTIONS *
1368              **************************/
1369
1370             r10              = _mm256_mul_ps(rsq10,rinv10);
1371             r10              = _mm256_andnot_ps(dummy_mask,r10);
1372
1373             /* Compute parameters for interactions between i and j atoms */
1374             qq10             = _mm256_mul_ps(iq1,jq0);
1375
1376             /* Calculate table index by multiplying r with table scale and truncate to integer */
1377             rt               = _mm256_mul_ps(r10,vftabscale);
1378             vfitab           = _mm256_cvttps_epi32(rt);
1379             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1380             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1381             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1382             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1383             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1384             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1385
1386             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1387             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1388                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1389             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1390                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1391             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1392                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1393             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1394                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1395             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1396             Heps             = _mm256_mul_ps(vfeps,H);
1397             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1398             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1399             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq10,FF),_mm256_mul_ps(vftabscale,rinv10)));
1400
1401             fscal            = felec;
1402
1403             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1404
1405             /* Calculate temporary vectorial force */
1406             tx               = _mm256_mul_ps(fscal,dx10);
1407             ty               = _mm256_mul_ps(fscal,dy10);
1408             tz               = _mm256_mul_ps(fscal,dz10);
1409
1410             /* Update vectorial force */
1411             fix1             = _mm256_add_ps(fix1,tx);
1412             fiy1             = _mm256_add_ps(fiy1,ty);
1413             fiz1             = _mm256_add_ps(fiz1,tz);
1414
1415             fjx0             = _mm256_add_ps(fjx0,tx);
1416             fjy0             = _mm256_add_ps(fjy0,ty);
1417             fjz0             = _mm256_add_ps(fjz0,tz);
1418
1419             /**************************
1420              * CALCULATE INTERACTIONS *
1421              **************************/
1422
1423             r20              = _mm256_mul_ps(rsq20,rinv20);
1424             r20              = _mm256_andnot_ps(dummy_mask,r20);
1425
1426             /* Compute parameters for interactions between i and j atoms */
1427             qq20             = _mm256_mul_ps(iq2,jq0);
1428
1429             /* Calculate table index by multiplying r with table scale and truncate to integer */
1430             rt               = _mm256_mul_ps(r20,vftabscale);
1431             vfitab           = _mm256_cvttps_epi32(rt);
1432             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1433             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1434             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1435             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1436             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1437             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1438
1439             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1440             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1441                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1442             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1443                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1444             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1445                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1446             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1447                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1448             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1449             Heps             = _mm256_mul_ps(vfeps,H);
1450             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1451             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1452             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq20,FF),_mm256_mul_ps(vftabscale,rinv20)));
1453
1454             fscal            = felec;
1455
1456             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1457
1458             /* Calculate temporary vectorial force */
1459             tx               = _mm256_mul_ps(fscal,dx20);
1460             ty               = _mm256_mul_ps(fscal,dy20);
1461             tz               = _mm256_mul_ps(fscal,dz20);
1462
1463             /* Update vectorial force */
1464             fix2             = _mm256_add_ps(fix2,tx);
1465             fiy2             = _mm256_add_ps(fiy2,ty);
1466             fiz2             = _mm256_add_ps(fiz2,tz);
1467
1468             fjx0             = _mm256_add_ps(fjx0,tx);
1469             fjy0             = _mm256_add_ps(fjy0,ty);
1470             fjz0             = _mm256_add_ps(fjz0,tz);
1471
1472             /**************************
1473              * CALCULATE INTERACTIONS *
1474              **************************/
1475
1476             r30              = _mm256_mul_ps(rsq30,rinv30);
1477             r30              = _mm256_andnot_ps(dummy_mask,r30);
1478
1479             /* Compute parameters for interactions between i and j atoms */
1480             qq30             = _mm256_mul_ps(iq3,jq0);
1481
1482             /* Calculate table index by multiplying r with table scale and truncate to integer */
1483             rt               = _mm256_mul_ps(r30,vftabscale);
1484             vfitab           = _mm256_cvttps_epi32(rt);
1485             vfeps            = _mm256_sub_ps(rt,_mm256_round_ps(rt, _MM_FROUND_FLOOR));
1486             /*         AVX1 does not support 256-bit integer operations, so now we go to 128-bit mode... */
1487             vfitab_lo        = _mm256_extractf128_si256(vfitab,0x0);
1488             vfitab_hi        = _mm256_extractf128_si256(vfitab,0x1);
1489             vfitab_lo        = _mm_slli_epi32(vfitab_lo,2);
1490             vfitab_hi        = _mm_slli_epi32(vfitab_hi,2);
1491
1492             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1493             Y                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,0)),
1494                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,0)));
1495             F                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,1)),
1496                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,1)));
1497             G                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,2)),
1498                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,2)));
1499             H                = gmx_mm256_set_m128(_mm_load_ps(vftab + _mm_extract_epi32(vfitab_hi,3)),
1500                                                   _mm_load_ps(vftab + _mm_extract_epi32(vfitab_lo,3)));
1501             GMX_MM256_HALFTRANSPOSE4_PS(Y,F,G,H);
1502             Heps             = _mm256_mul_ps(vfeps,H);
1503             Fp               = _mm256_add_ps(F,_mm256_mul_ps(vfeps,_mm256_add_ps(G,Heps)));
1504             FF               = _mm256_add_ps(Fp,_mm256_mul_ps(vfeps,_mm256_add_ps(G,_mm256_add_ps(Heps,Heps))));
1505             felec            = _mm256_xor_ps(signbit,_mm256_mul_ps(_mm256_mul_ps(qq30,FF),_mm256_mul_ps(vftabscale,rinv30)));
1506
1507             fscal            = felec;
1508
1509             fscal            = _mm256_andnot_ps(dummy_mask,fscal);
1510
1511             /* Calculate temporary vectorial force */
1512             tx               = _mm256_mul_ps(fscal,dx30);
1513             ty               = _mm256_mul_ps(fscal,dy30);
1514             tz               = _mm256_mul_ps(fscal,dz30);
1515
1516             /* Update vectorial force */
1517             fix3             = _mm256_add_ps(fix3,tx);
1518             fiy3             = _mm256_add_ps(fiy3,ty);
1519             fiz3             = _mm256_add_ps(fiz3,tz);
1520
1521             fjx0             = _mm256_add_ps(fjx0,tx);
1522             fjy0             = _mm256_add_ps(fjy0,ty);
1523             fjz0             = _mm256_add_ps(fjz0,tz);
1524
1525             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1526             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1527             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1528             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1529             fjptrE             = (jnrlistE>=0) ? f+j_coord_offsetE : scratch;
1530             fjptrF             = (jnrlistF>=0) ? f+j_coord_offsetF : scratch;
1531             fjptrG             = (jnrlistG>=0) ? f+j_coord_offsetG : scratch;
1532             fjptrH             = (jnrlistH>=0) ? f+j_coord_offsetH : scratch;
1533
1534             gmx_mm256_decrement_1rvec_8ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjptrE,fjptrF,fjptrG,fjptrH,fjx0,fjy0,fjz0);
1535
1536             /* Inner loop uses 150 flops */
1537         }
1538
1539         /* End of innermost loop */
1540
1541         gmx_mm256_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1542                                                  f+i_coord_offset,fshift+i_shift_offset);
1543
1544         /* Increment number of inner iterations */
1545         inneriter                  += j_index_end - j_index_start;
1546
1547         /* Outer loop uses 24 flops */
1548     }
1549
1550     /* Increment number of outer iterations */
1551     outeriter        += nri;
1552
1553     /* Update outer/inner flops */
1554
1555     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*150);
1556 }