src/gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_single/nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_sse4_1_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS sse4_1_single kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_sse4_1_single.h"
  50 #include "kernelutil_x86_sse4_1_single.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_VF_sse4_1_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            LJEwald
  56  * Geometry:                   Water4-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_VF_sse4_1_single
  61                     (t_nblist                    * gmx_restrict       nlist,
  62                      rvec                        * gmx_restrict          xx,
  63                      rvec                        * gmx_restrict          ff,
  64                      t_forcerec                  * gmx_restrict          fr,
  65                      t_mdatoms                   * gmx_restrict     mdatoms,
  66                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  67                      t_nrnb                      * gmx_restrict        nrnb)
  68 {
  69     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  70      * just 0 for non-waters.
  71      * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
  72      * jnr indices corresponding to data put in the four positions in the SIMD register.
  73      */
  74     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  75     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  76     int              jnrA,jnrB,jnrC,jnrD;
  77     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  78     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  79     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  80     real             rcutoff_scalar;
  81     real             *shiftvec,*fshift,*x,*f;
  82     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  83     real             scratch[4*DIM];
  84     __m128           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  85     int              vdwioffset0;
  86     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  87     int              vdwioffset1;
  88     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  89     int              vdwioffset2;
  90     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  91     int              vdwioffset3;
  92     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  93     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  94     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  95     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  96     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  97     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  98     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
  99     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 100     real             *charge;
 101     int              nvdwtype;
 102     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 103     int              *vdwtype;
 104     real             *vdwparam;
 105     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 106     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 107     __m128           c6grid_00;
 108     __m128           c6grid_10;
 109     __m128           c6grid_20;
 110     __m128           c6grid_30;
 111     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 112     real             *vdwgridparam;
 113     __m128           one_half  = _mm_set1_ps(0.5);
 114     __m128           minus_one = _mm_set1_ps(-1.0);
 115     __m128i          ewitab;
 116     __m128           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 117     real             *ewtab;
 118     __m128           dummy_mask,cutoff_mask;
 119     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 120     __m128           one     = _mm_set1_ps(1.0);
 121     __m128           two     = _mm_set1_ps(2.0);
 122     x                = xx[0];
 123     f                = ff[0];
 124
 125     nri              = nlist->nri;
 126     iinr             = nlist->iinr;
 127     jindex           = nlist->jindex;
 128     jjnr             = nlist->jjnr;
 129     shiftidx         = nlist->shift;
 130     gid              = nlist->gid;
 131     shiftvec         = fr->shift_vec[0];
 132     fshift           = fr->fshift[0];
 133     facel            = _mm_set1_ps(fr->epsfac);
 134     charge           = mdatoms->chargeA;
 135     nvdwtype         = fr->ntype;
 136     vdwparam         = fr->nbfp;
 137     vdwtype          = mdatoms->typeA;
 138     vdwgridparam     = fr->ljpme_c6grid;
 139     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
 140     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
 141     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
 142
 143     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 144     ewtab            = fr->ic->tabq_coul_FDV0;
 145     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 146     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 147
 148     /* Setup water-specific parameters */
 149     inr              = nlist->iinr[0];
 150     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 151     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 152     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 153     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 154
 155     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 156     rcutoff_scalar   = fr->rcoulomb;
 157     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 158     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 159
 160     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
 161     rvdw             = _mm_set1_ps(fr->rvdw);
 162
 163     /* Avoid stupid compiler warnings */
 164     jnrA = jnrB = jnrC = jnrD = 0;
 165     j_coord_offsetA = 0;
 166     j_coord_offsetB = 0;
 167     j_coord_offsetC = 0;
 168     j_coord_offsetD = 0;
 169
 170     outeriter        = 0;
 171     inneriter        = 0;
 172
 173     for(iidx=0;iidx<4*DIM;iidx++)
 174     {
 175         scratch[iidx] = 0.0;
 176     }
 177
 178     /* Start outer loop over neighborlists */
 179     for(iidx=0; iidx<nri; iidx++)
 180     {
 181         /* Load shift vector for this list */
 182         i_shift_offset   = DIM*shiftidx[iidx];
 183
 184         /* Load limits for loop over neighbors */
 185         j_index_start    = jindex[iidx];
 186         j_index_end      = jindex[iidx+1];
 187
 188         /* Get outer coordinate index */
 189         inr              = iinr[iidx];
 190         i_coord_offset   = DIM*inr;
 191
 192         /* Load i particle coords and add shift vector */
 193         gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 194                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 195
 196         fix0             = _mm_setzero_ps();
 197         fiy0             = _mm_setzero_ps();
 198         fiz0             = _mm_setzero_ps();
 199         fix1             = _mm_setzero_ps();
 200         fiy1             = _mm_setzero_ps();
 201         fiz1             = _mm_setzero_ps();
 202         fix2             = _mm_setzero_ps();
 203         fiy2             = _mm_setzero_ps();
 204         fiz2             = _mm_setzero_ps();
 205         fix3             = _mm_setzero_ps();
 206         fiy3             = _mm_setzero_ps();
 207         fiz3             = _mm_setzero_ps();
 208
 209         /* Reset potential sums */
 210         velecsum         = _mm_setzero_ps();
 211         vvdwsum          = _mm_setzero_ps();
 212
 213         /* Start inner kernel loop */
 214         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 215         {
 216
 217             /* Get j neighbor index, and coordinate index */
 218             jnrA             = jjnr[jidx];
 219             jnrB             = jjnr[jidx+1];
 220             jnrC             = jjnr[jidx+2];
 221             jnrD             = jjnr[jidx+3];
 222             j_coord_offsetA  = DIM*jnrA;
 223             j_coord_offsetB  = DIM*jnrB;
 224             j_coord_offsetC  = DIM*jnrC;
 225             j_coord_offsetD  = DIM*jnrD;
 226
 227             /* load j atom coordinates */
 228             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 229                                               x+j_coord_offsetC,x+j_coord_offsetD,
 230                                               &jx0,&jy0,&jz0);
 231
 232             /* Calculate displacement vector */
 233             dx00             = _mm_sub_ps(ix0,jx0);
 234             dy00             = _mm_sub_ps(iy0,jy0);
 235             dz00             = _mm_sub_ps(iz0,jz0);
 236             dx10             = _mm_sub_ps(ix1,jx0);
 237             dy10             = _mm_sub_ps(iy1,jy0);
 238             dz10             = _mm_sub_ps(iz1,jz0);
 239             dx20             = _mm_sub_ps(ix2,jx0);
 240             dy20             = _mm_sub_ps(iy2,jy0);
 241             dz20             = _mm_sub_ps(iz2,jz0);
 242             dx30             = _mm_sub_ps(ix3,jx0);
 243             dy30             = _mm_sub_ps(iy3,jy0);
 244             dz30             = _mm_sub_ps(iz3,jz0);
 245
 246             /* Calculate squared distance and things based on it */
 247             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 248             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 249             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 250             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 251
 252             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 253             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 254             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 255             rinv30           = gmx_mm_invsqrt_ps(rsq30);
 256
 257             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 258             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 259             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 260             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 261
 262             /* Load parameters for j particles */
 263             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 264                                                               charge+jnrC+0,charge+jnrD+0);
 265             vdwjidx0A        = 2*vdwtype[jnrA+0];
 266             vdwjidx0B        = 2*vdwtype[jnrB+0];
 267             vdwjidx0C        = 2*vdwtype[jnrC+0];
 268             vdwjidx0D        = 2*vdwtype[jnrD+0];
 269
 270             fjx0             = _mm_setzero_ps();
 271             fjy0             = _mm_setzero_ps();
 272             fjz0             = _mm_setzero_ps();
 273
 274             /**************************
 275              * CALCULATE INTERACTIONS *
 276              **************************/
 277
 278             if (gmx_mm_any_lt(rsq00,rcutoff2))
 279             {
 280
 281             r00              = _mm_mul_ps(rsq00,rinv00);
 282
 283             /* Compute parameters for interactions between i and j atoms */
 284             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 285                                          vdwparam+vdwioffset0+vdwjidx0B,
 286                                          vdwparam+vdwioffset0+vdwjidx0C,
 287                                          vdwparam+vdwioffset0+vdwjidx0D,
 288                                          &c6_00,&c12_00);
 289
 290             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 291                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 292                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 293                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 294
 295             /* Analytical LJ-PME */
 296             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 297             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 298             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 299             exponent         = gmx_simd_exp_r(ewcljrsq);
 300             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 301             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 302             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 303             vvdw6            = _mm_mul_ps(_mm_sub_ps(c6_00,_mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly))),rinvsix);
 304             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 305             vvdw             = _mm_sub_ps(_mm_mul_ps( _mm_sub_ps(vvdw12 , _mm_mul_ps(c12_00,_mm_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))),one_twelfth),
 306                                _mm_mul_ps( _mm_sub_ps(vvdw6,_mm_add_ps(_mm_mul_ps(c6_00,sh_vdw_invrcut6),_mm_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
 307             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 308             fvdw             = _mm_mul_ps(_mm_sub_ps(vvdw12,_mm_sub_ps(vvdw6,_mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6)))),rinvsq00);
 309
 310             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 311
 312             /* Update potential sum for this i atom from the interaction with this j atom. */
 313             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 314             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 315
 316             fscal            = fvdw;
 317
 318             fscal            = _mm_and_ps(fscal,cutoff_mask);
 319
 320             /* Calculate temporary vectorial force */
 321             tx               = _mm_mul_ps(fscal,dx00);
 322             ty               = _mm_mul_ps(fscal,dy00);
 323             tz               = _mm_mul_ps(fscal,dz00);
 324
 325             /* Update vectorial force */
 326             fix0             = _mm_add_ps(fix0,tx);
 327             fiy0             = _mm_add_ps(fiy0,ty);
 328             fiz0             = _mm_add_ps(fiz0,tz);
 329
 330             fjx0             = _mm_add_ps(fjx0,tx);
 331             fjy0             = _mm_add_ps(fjy0,ty);
 332             fjz0             = _mm_add_ps(fjz0,tz);
 333
 334             }
 335
 336             /**************************
 337              * CALCULATE INTERACTIONS *
 338              **************************/
 339
 340             if (gmx_mm_any_lt(rsq10,rcutoff2))
 341             {
 342
 343             r10              = _mm_mul_ps(rsq10,rinv10);
 344
 345             /* Compute parameters for interactions between i and j atoms */
 346             qq10             = _mm_mul_ps(iq1,jq0);
 347
 348             /* EWALD ELECTROSTATICS */
 349
 350             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 351             ewrt             = _mm_mul_ps(r10,ewtabscale);
 352             ewitab           = _mm_cvttps_epi32(ewrt);
 353             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 354             ewitab           = _mm_slli_epi32(ewitab,2);
 355             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 356             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 357             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 358             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 359             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 360             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 361             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 362             velec            = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec));
 363             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
 364
 365             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 366
 367             /* Update potential sum for this i atom from the interaction with this j atom. */
 368             velec            = _mm_and_ps(velec,cutoff_mask);
 369             velecsum         = _mm_add_ps(velecsum,velec);
 370
 371             fscal            = felec;
 372
 373             fscal            = _mm_and_ps(fscal,cutoff_mask);
 374
 375             /* Calculate temporary vectorial force */
 376             tx               = _mm_mul_ps(fscal,dx10);
 377             ty               = _mm_mul_ps(fscal,dy10);
 378             tz               = _mm_mul_ps(fscal,dz10);
 379
 380             /* Update vectorial force */
 381             fix1             = _mm_add_ps(fix1,tx);
 382             fiy1             = _mm_add_ps(fiy1,ty);
 383             fiz1             = _mm_add_ps(fiz1,tz);
 384
 385             fjx0             = _mm_add_ps(fjx0,tx);
 386             fjy0             = _mm_add_ps(fjy0,ty);
 387             fjz0             = _mm_add_ps(fjz0,tz);
 388
 389             }
 390
 391             /**************************
 392              * CALCULATE INTERACTIONS *
 393              **************************/
 394
 395             if (gmx_mm_any_lt(rsq20,rcutoff2))
 396             {
 397
 398             r20              = _mm_mul_ps(rsq20,rinv20);
 399
 400             /* Compute parameters for interactions between i and j atoms */
 401             qq20             = _mm_mul_ps(iq2,jq0);
 402
 403             /* EWALD ELECTROSTATICS */
 404
 405             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 406             ewrt             = _mm_mul_ps(r20,ewtabscale);
 407             ewitab           = _mm_cvttps_epi32(ewrt);
 408             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 409             ewitab           = _mm_slli_epi32(ewitab,2);
 410             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 411             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 412             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 413             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 414             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 415             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 416             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 417             velec            = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec));
 418             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
 419
 420             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 421
 422             /* Update potential sum for this i atom from the interaction with this j atom. */
 423             velec            = _mm_and_ps(velec,cutoff_mask);
 424             velecsum         = _mm_add_ps(velecsum,velec);
 425
 426             fscal            = felec;
 427
 428             fscal            = _mm_and_ps(fscal,cutoff_mask);
 429
 430             /* Calculate temporary vectorial force */
 431             tx               = _mm_mul_ps(fscal,dx20);
 432             ty               = _mm_mul_ps(fscal,dy20);
 433             tz               = _mm_mul_ps(fscal,dz20);
 434
 435             /* Update vectorial force */
 436             fix2             = _mm_add_ps(fix2,tx);
 437             fiy2             = _mm_add_ps(fiy2,ty);
 438             fiz2             = _mm_add_ps(fiz2,tz);
 439
 440             fjx0             = _mm_add_ps(fjx0,tx);
 441             fjy0             = _mm_add_ps(fjy0,ty);
 442             fjz0             = _mm_add_ps(fjz0,tz);
 443
 444             }
 445
 446             /**************************
 447              * CALCULATE INTERACTIONS *
 448              **************************/
 449
 450             if (gmx_mm_any_lt(rsq30,rcutoff2))
 451             {
 452
 453             r30              = _mm_mul_ps(rsq30,rinv30);
 454
 455             /* Compute parameters for interactions between i and j atoms */
 456             qq30             = _mm_mul_ps(iq3,jq0);
 457
 458             /* EWALD ELECTROSTATICS */
 459
 460             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 461             ewrt             = _mm_mul_ps(r30,ewtabscale);
 462             ewitab           = _mm_cvttps_epi32(ewrt);
 463             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 464             ewitab           = _mm_slli_epi32(ewitab,2);
 465             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 466             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 467             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 468             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 469             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 470             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 471             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 472             velec            = _mm_mul_ps(qq30,_mm_sub_ps(_mm_sub_ps(rinv30,sh_ewald),velec));
 473             felec            = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
 474
 475             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
 476
 477             /* Update potential sum for this i atom from the interaction with this j atom. */
 478             velec            = _mm_and_ps(velec,cutoff_mask);
 479             velecsum         = _mm_add_ps(velecsum,velec);
 480
 481             fscal            = felec;
 482
 483             fscal            = _mm_and_ps(fscal,cutoff_mask);
 484
 485             /* Calculate temporary vectorial force */
 486             tx               = _mm_mul_ps(fscal,dx30);
 487             ty               = _mm_mul_ps(fscal,dy30);
 488             tz               = _mm_mul_ps(fscal,dz30);
 489
 490             /* Update vectorial force */
 491             fix3             = _mm_add_ps(fix3,tx);
 492             fiy3             = _mm_add_ps(fiy3,ty);
 493             fiz3             = _mm_add_ps(fiz3,tz);
 494
 495             fjx0             = _mm_add_ps(fjx0,tx);
 496             fjy0             = _mm_add_ps(fjy0,ty);
 497             fjz0             = _mm_add_ps(fjz0,tz);
 498
 499             }
 500
 501             fjptrA             = f+j_coord_offsetA;
 502             fjptrB             = f+j_coord_offsetB;
 503             fjptrC             = f+j_coord_offsetC;
 504             fjptrD             = f+j_coord_offsetD;
 505
 506             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 507
 508             /* Inner loop uses 200 flops */
 509         }
 510
 511         if(jidx<j_index_end)
 512         {
 513
 514             /* Get j neighbor index, and coordinate index */
 515             jnrlistA         = jjnr[jidx];
 516             jnrlistB         = jjnr[jidx+1];
 517             jnrlistC         = jjnr[jidx+2];
 518             jnrlistD         = jjnr[jidx+3];
 519             /* Sign of each element will be negative for non-real atoms.
 520              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 521              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 522              */
 523             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 524             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 525             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 526             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 527             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 528             j_coord_offsetA  = DIM*jnrA;
 529             j_coord_offsetB  = DIM*jnrB;
 530             j_coord_offsetC  = DIM*jnrC;
 531             j_coord_offsetD  = DIM*jnrD;
 532
 533             /* load j atom coordinates */
 534             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 535                                               x+j_coord_offsetC,x+j_coord_offsetD,
 536                                               &jx0,&jy0,&jz0);
 537
 538             /* Calculate displacement vector */
 539             dx00             = _mm_sub_ps(ix0,jx0);
 540             dy00             = _mm_sub_ps(iy0,jy0);
 541             dz00             = _mm_sub_ps(iz0,jz0);
 542             dx10             = _mm_sub_ps(ix1,jx0);
 543             dy10             = _mm_sub_ps(iy1,jy0);
 544             dz10             = _mm_sub_ps(iz1,jz0);
 545             dx20             = _mm_sub_ps(ix2,jx0);
 546             dy20             = _mm_sub_ps(iy2,jy0);
 547             dz20             = _mm_sub_ps(iz2,jz0);
 548             dx30             = _mm_sub_ps(ix3,jx0);
 549             dy30             = _mm_sub_ps(iy3,jy0);
 550             dz30             = _mm_sub_ps(iz3,jz0);
 551
 552             /* Calculate squared distance and things based on it */
 553             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 554             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 555             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 556             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 557
 558             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 559             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 560             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 561             rinv30           = gmx_mm_invsqrt_ps(rsq30);
 562
 563             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 564             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 565             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 566             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 567
 568             /* Load parameters for j particles */
 569             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 570                                                               charge+jnrC+0,charge+jnrD+0);
 571             vdwjidx0A        = 2*vdwtype[jnrA+0];
 572             vdwjidx0B        = 2*vdwtype[jnrB+0];
 573             vdwjidx0C        = 2*vdwtype[jnrC+0];
 574             vdwjidx0D        = 2*vdwtype[jnrD+0];
 575
 576             fjx0             = _mm_setzero_ps();
 577             fjy0             = _mm_setzero_ps();
 578             fjz0             = _mm_setzero_ps();
 579
 580             /**************************
 581              * CALCULATE INTERACTIONS *
 582              **************************/
 583
 584             if (gmx_mm_any_lt(rsq00,rcutoff2))
 585             {
 586
 587             r00              = _mm_mul_ps(rsq00,rinv00);
 588             r00              = _mm_andnot_ps(dummy_mask,r00);
 589
 590             /* Compute parameters for interactions between i and j atoms */
 591             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 592                                          vdwparam+vdwioffset0+vdwjidx0B,
 593                                          vdwparam+vdwioffset0+vdwjidx0C,
 594                                          vdwparam+vdwioffset0+vdwjidx0D,
 595                                          &c6_00,&c12_00);
 596
 597             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 598                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 599                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 600                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 601
 602             /* Analytical LJ-PME */
 603             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 604             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 605             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 606             exponent         = gmx_simd_exp_r(ewcljrsq);
 607             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 608             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 609             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 610             vvdw6            = _mm_mul_ps(_mm_sub_ps(c6_00,_mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly))),rinvsix);
 611             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 612             vvdw             = _mm_sub_ps(_mm_mul_ps( _mm_sub_ps(vvdw12 , _mm_mul_ps(c12_00,_mm_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6))),one_twelfth),
 613                                _mm_mul_ps( _mm_sub_ps(vvdw6,_mm_add_ps(_mm_mul_ps(c6_00,sh_vdw_invrcut6),_mm_mul_ps(c6grid_00,sh_lj_ewald))),one_sixth));
 614             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 615             fvdw             = _mm_mul_ps(_mm_sub_ps(vvdw12,_mm_sub_ps(vvdw6,_mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6)))),rinvsq00);
 616
 617             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 618
 619             /* Update potential sum for this i atom from the interaction with this j atom. */
 620             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 621             vvdw             = _mm_andnot_ps(dummy_mask,vvdw);
 622             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 623
 624             fscal            = fvdw;
 625
 626             fscal            = _mm_and_ps(fscal,cutoff_mask);
 627
 628             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 629
 630             /* Calculate temporary vectorial force */
 631             tx               = _mm_mul_ps(fscal,dx00);
 632             ty               = _mm_mul_ps(fscal,dy00);
 633             tz               = _mm_mul_ps(fscal,dz00);
 634
 635             /* Update vectorial force */
 636             fix0             = _mm_add_ps(fix0,tx);
 637             fiy0             = _mm_add_ps(fiy0,ty);
 638             fiz0             = _mm_add_ps(fiz0,tz);
 639
 640             fjx0             = _mm_add_ps(fjx0,tx);
 641             fjy0             = _mm_add_ps(fjy0,ty);
 642             fjz0             = _mm_add_ps(fjz0,tz);
 643
 644             }
 645
 646             /**************************
 647              * CALCULATE INTERACTIONS *
 648              **************************/
 649
 650             if (gmx_mm_any_lt(rsq10,rcutoff2))
 651             {
 652
 653             r10              = _mm_mul_ps(rsq10,rinv10);
 654             r10              = _mm_andnot_ps(dummy_mask,r10);
 655
 656             /* Compute parameters for interactions between i and j atoms */
 657             qq10             = _mm_mul_ps(iq1,jq0);
 658
 659             /* EWALD ELECTROSTATICS */
 660
 661             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 662             ewrt             = _mm_mul_ps(r10,ewtabscale);
 663             ewitab           = _mm_cvttps_epi32(ewrt);
 664             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 665             ewitab           = _mm_slli_epi32(ewitab,2);
 666             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 667             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 668             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 669             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 670             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 671             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 672             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 673             velec            = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec));
 674             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
 675
 676             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 677
 678             /* Update potential sum for this i atom from the interaction with this j atom. */
 679             velec            = _mm_and_ps(velec,cutoff_mask);
 680             velec            = _mm_andnot_ps(dummy_mask,velec);
 681             velecsum         = _mm_add_ps(velecsum,velec);
 682
 683             fscal            = felec;
 684
 685             fscal            = _mm_and_ps(fscal,cutoff_mask);
 686
 687             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 688
 689             /* Calculate temporary vectorial force */
 690             tx               = _mm_mul_ps(fscal,dx10);
 691             ty               = _mm_mul_ps(fscal,dy10);
 692             tz               = _mm_mul_ps(fscal,dz10);
 693
 694             /* Update vectorial force */
 695             fix1             = _mm_add_ps(fix1,tx);
 696             fiy1             = _mm_add_ps(fiy1,ty);
 697             fiz1             = _mm_add_ps(fiz1,tz);
 698
 699             fjx0             = _mm_add_ps(fjx0,tx);
 700             fjy0             = _mm_add_ps(fjy0,ty);
 701             fjz0             = _mm_add_ps(fjz0,tz);
 702
 703             }
 704
 705             /**************************
 706              * CALCULATE INTERACTIONS *
 707              **************************/
 708
 709             if (gmx_mm_any_lt(rsq20,rcutoff2))
 710             {
 711
 712             r20              = _mm_mul_ps(rsq20,rinv20);
 713             r20              = _mm_andnot_ps(dummy_mask,r20);
 714
 715             /* Compute parameters for interactions between i and j atoms */
 716             qq20             = _mm_mul_ps(iq2,jq0);
 717
 718             /* EWALD ELECTROSTATICS */
 719
 720             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 721             ewrt             = _mm_mul_ps(r20,ewtabscale);
 722             ewitab           = _mm_cvttps_epi32(ewrt);
 723             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 724             ewitab           = _mm_slli_epi32(ewitab,2);
 725             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 726             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 727             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 728             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 729             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 730             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 731             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 732             velec            = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec));
 733             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
 734
 735             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 736
 737             /* Update potential sum for this i atom from the interaction with this j atom. */
 738             velec            = _mm_and_ps(velec,cutoff_mask);
 739             velec            = _mm_andnot_ps(dummy_mask,velec);
 740             velecsum         = _mm_add_ps(velecsum,velec);
 741
 742             fscal            = felec;
 743
 744             fscal            = _mm_and_ps(fscal,cutoff_mask);
 745
 746             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 747
 748             /* Calculate temporary vectorial force */
 749             tx               = _mm_mul_ps(fscal,dx20);
 750             ty               = _mm_mul_ps(fscal,dy20);
 751             tz               = _mm_mul_ps(fscal,dz20);
 752
 753             /* Update vectorial force */
 754             fix2             = _mm_add_ps(fix2,tx);
 755             fiy2             = _mm_add_ps(fiy2,ty);
 756             fiz2             = _mm_add_ps(fiz2,tz);
 757
 758             fjx0             = _mm_add_ps(fjx0,tx);
 759             fjy0             = _mm_add_ps(fjy0,ty);
 760             fjz0             = _mm_add_ps(fjz0,tz);
 761
 762             }
 763
 764             /**************************
 765              * CALCULATE INTERACTIONS *
 766              **************************/
 767
 768             if (gmx_mm_any_lt(rsq30,rcutoff2))
 769             {
 770
 771             r30              = _mm_mul_ps(rsq30,rinv30);
 772             r30              = _mm_andnot_ps(dummy_mask,r30);
 773
 774             /* Compute parameters for interactions between i and j atoms */
 775             qq30             = _mm_mul_ps(iq3,jq0);
 776
 777             /* EWALD ELECTROSTATICS */
 778
 779             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 780             ewrt             = _mm_mul_ps(r30,ewtabscale);
 781             ewitab           = _mm_cvttps_epi32(ewrt);
 782             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 783             ewitab           = _mm_slli_epi32(ewitab,2);
 784             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 785             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 786             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 787             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 788             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 789             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 790             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 791             velec            = _mm_mul_ps(qq30,_mm_sub_ps(_mm_sub_ps(rinv30,sh_ewald),velec));
 792             felec            = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
 793
 794             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
 795
 796             /* Update potential sum for this i atom from the interaction with this j atom. */
 797             velec            = _mm_and_ps(velec,cutoff_mask);
 798             velec            = _mm_andnot_ps(dummy_mask,velec);
 799             velecsum         = _mm_add_ps(velecsum,velec);
 800
 801             fscal            = felec;
 802
 803             fscal            = _mm_and_ps(fscal,cutoff_mask);
 804
 805             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 806
 807             /* Calculate temporary vectorial force */
 808             tx               = _mm_mul_ps(fscal,dx30);
 809             ty               = _mm_mul_ps(fscal,dy30);
 810             tz               = _mm_mul_ps(fscal,dz30);
 811
 812             /* Update vectorial force */
 813             fix3             = _mm_add_ps(fix3,tx);
 814             fiy3             = _mm_add_ps(fiy3,ty);
 815             fiz3             = _mm_add_ps(fiz3,tz);
 816
 817             fjx0             = _mm_add_ps(fjx0,tx);
 818             fjy0             = _mm_add_ps(fjy0,ty);
 819             fjz0             = _mm_add_ps(fjz0,tz);
 820
 821             }
 822
 823             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 824             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 825             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 826             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 827
 828             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 829
 830             /* Inner loop uses 204 flops */
 831         }
 832
 833         /* End of innermost loop */
 834
 835         gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 836                                               f+i_coord_offset,fshift+i_shift_offset);
 837
 838         ggid                        = gid[iidx];
 839         /* Update potential energies */
 840         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 841         gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 842
 843         /* Increment number of inner iterations */
 844         inneriter                  += j_index_end - j_index_start;
 845
 846         /* Outer loop uses 26 flops */
 847     }
 848
 849     /* Increment number of outer iterations */
 850     outeriter        += nri;
 851
 852     /* Update outer/inner flops */
 853
 854     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*204);
 855 }
 856 /*
 857  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_F_sse4_1_single
 858  * Electrostatics interaction: Ewald
 859  * VdW interaction:            LJEwald
 860  * Geometry:                   Water4-Particle
 861  * Calculate force/pot:        Force
 862  */
 863 void
 864 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW4P1_F_sse4_1_single
 865                     (t_nblist                    * gmx_restrict       nlist,
 866                      rvec                        * gmx_restrict          xx,
 867                      rvec                        * gmx_restrict          ff,
 868                      t_forcerec                  * gmx_restrict          fr,
 869                      t_mdatoms                   * gmx_restrict     mdatoms,
 870                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 871                      t_nrnb                      * gmx_restrict        nrnb)
 872 {
 873     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 874      * just 0 for non-waters.
 875      * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
 876      * jnr indices corresponding to data put in the four positions in the SIMD register.
 877      */
 878     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 879     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 880     int              jnrA,jnrB,jnrC,jnrD;
 881     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 882     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 883     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 884     real             rcutoff_scalar;
 885     real             *shiftvec,*fshift,*x,*f;
 886     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 887     real             scratch[4*DIM];
 888     __m128           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 889     int              vdwioffset0;
 890     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 891     int              vdwioffset1;
 892     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 893     int              vdwioffset2;
 894     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 895     int              vdwioffset3;
 896     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 897     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 898     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 899     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 900     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 901     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 902     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 903     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 904     real             *charge;
 905     int              nvdwtype;
 906     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 907     int              *vdwtype;
 908     real             *vdwparam;
 909     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 910     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 911     __m128           c6grid_00;
 912     __m128           c6grid_10;
 913     __m128           c6grid_20;
 914     __m128           c6grid_30;
 915     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 916     real             *vdwgridparam;
 917     __m128           one_half  = _mm_set1_ps(0.5);
 918     __m128           minus_one = _mm_set1_ps(-1.0);
 919     __m128i          ewitab;
 920     __m128           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 921     real             *ewtab;
 922     __m128           dummy_mask,cutoff_mask;
 923     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 924     __m128           one     = _mm_set1_ps(1.0);
 925     __m128           two     = _mm_set1_ps(2.0);
 926     x                = xx[0];
 927     f                = ff[0];
 928
 929     nri              = nlist->nri;
 930     iinr             = nlist->iinr;
 931     jindex           = nlist->jindex;
 932     jjnr             = nlist->jjnr;
 933     shiftidx         = nlist->shift;
 934     gid              = nlist->gid;
 935     shiftvec         = fr->shift_vec[0];
 936     fshift           = fr->fshift[0];
 937     facel            = _mm_set1_ps(fr->epsfac);
 938     charge           = mdatoms->chargeA;
 939     nvdwtype         = fr->ntype;
 940     vdwparam         = fr->nbfp;
 941     vdwtype          = mdatoms->typeA;
 942     vdwgridparam     = fr->ljpme_c6grid;
 943     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
 944     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
 945     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
 946
 947     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 948     ewtab            = fr->ic->tabq_coul_F;
 949     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 950     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 951
 952     /* Setup water-specific parameters */
 953     inr              = nlist->iinr[0];
 954     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 955     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 956     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 957     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 958
 959     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 960     rcutoff_scalar   = fr->rcoulomb;
 961     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 962     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 963
 964     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
 965     rvdw             = _mm_set1_ps(fr->rvdw);
 966
 967     /* Avoid stupid compiler warnings */
 968     jnrA = jnrB = jnrC = jnrD = 0;
 969     j_coord_offsetA = 0;
 970     j_coord_offsetB = 0;
 971     j_coord_offsetC = 0;
 972     j_coord_offsetD = 0;
 973
 974     outeriter        = 0;
 975     inneriter        = 0;
 976
 977     for(iidx=0;iidx<4*DIM;iidx++)
 978     {
 979         scratch[iidx] = 0.0;
 980     }
 981
 982     /* Start outer loop over neighborlists */
 983     for(iidx=0; iidx<nri; iidx++)
 984     {
 985         /* Load shift vector for this list */
 986         i_shift_offset   = DIM*shiftidx[iidx];
 987
 988         /* Load limits for loop over neighbors */
 989         j_index_start    = jindex[iidx];
 990         j_index_end      = jindex[iidx+1];
 991
 992         /* Get outer coordinate index */
 993         inr              = iinr[iidx];
 994         i_coord_offset   = DIM*inr;
 995
 996         /* Load i particle coords and add shift vector */
 997         gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 998                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 999
1000         fix0             = _mm_setzero_ps();
1001         fiy0             = _mm_setzero_ps();
1002         fiz0             = _mm_setzero_ps();
1003         fix1             = _mm_setzero_ps();
1004         fiy1             = _mm_setzero_ps();
1005         fiz1             = _mm_setzero_ps();
1006         fix2             = _mm_setzero_ps();
1007         fiy2             = _mm_setzero_ps();
1008         fiz2             = _mm_setzero_ps();
1009         fix3             = _mm_setzero_ps();
1010         fiy3             = _mm_setzero_ps();
1011         fiz3             = _mm_setzero_ps();
1012
1013         /* Start inner kernel loop */
1014         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1015         {
1016
1017             /* Get j neighbor index, and coordinate index */
1018             jnrA             = jjnr[jidx];
1019             jnrB             = jjnr[jidx+1];
1020             jnrC             = jjnr[jidx+2];
1021             jnrD             = jjnr[jidx+3];
1022             j_coord_offsetA  = DIM*jnrA;
1023             j_coord_offsetB  = DIM*jnrB;
1024             j_coord_offsetC  = DIM*jnrC;
1025             j_coord_offsetD  = DIM*jnrD;
1026
1027             /* load j atom coordinates */
1028             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1029                                               x+j_coord_offsetC,x+j_coord_offsetD,
1030                                               &jx0,&jy0,&jz0);
1031
1032             /* Calculate displacement vector */
1033             dx00             = _mm_sub_ps(ix0,jx0);
1034             dy00             = _mm_sub_ps(iy0,jy0);
1035             dz00             = _mm_sub_ps(iz0,jz0);
1036             dx10             = _mm_sub_ps(ix1,jx0);
1037             dy10             = _mm_sub_ps(iy1,jy0);
1038             dz10             = _mm_sub_ps(iz1,jz0);
1039             dx20             = _mm_sub_ps(ix2,jx0);
1040             dy20             = _mm_sub_ps(iy2,jy0);
1041             dz20             = _mm_sub_ps(iz2,jz0);
1042             dx30             = _mm_sub_ps(ix3,jx0);
1043             dy30             = _mm_sub_ps(iy3,jy0);
1044             dz30             = _mm_sub_ps(iz3,jz0);
1045
1046             /* Calculate squared distance and things based on it */
1047             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1048             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1049             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1050             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1051
1052             rinv00           = gmx_mm_invsqrt_ps(rsq00);
1053             rinv10           = gmx_mm_invsqrt_ps(rsq10);
1054             rinv20           = gmx_mm_invsqrt_ps(rsq20);
1055             rinv30           = gmx_mm_invsqrt_ps(rsq30);
1056
1057             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1058             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1059             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1060             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
1061
1062             /* Load parameters for j particles */
1063             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1064                                                               charge+jnrC+0,charge+jnrD+0);
1065             vdwjidx0A        = 2*vdwtype[jnrA+0];
1066             vdwjidx0B        = 2*vdwtype[jnrB+0];
1067             vdwjidx0C        = 2*vdwtype[jnrC+0];
1068             vdwjidx0D        = 2*vdwtype[jnrD+0];
1069
1070             fjx0             = _mm_setzero_ps();
1071             fjy0             = _mm_setzero_ps();
1072             fjz0             = _mm_setzero_ps();
1073
1074             /**************************
1075              * CALCULATE INTERACTIONS *
1076              **************************/
1077
1078             if (gmx_mm_any_lt(rsq00,rcutoff2))
1079             {
1080
1081             r00              = _mm_mul_ps(rsq00,rinv00);
1082
1083             /* Compute parameters for interactions between i and j atoms */
1084             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1085                                          vdwparam+vdwioffset0+vdwjidx0B,
1086                                          vdwparam+vdwioffset0+vdwjidx0C,
1087                                          vdwparam+vdwioffset0+vdwjidx0D,
1088                                          &c6_00,&c12_00);
1089
1090             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
1091                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
1092                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
1093                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
1094
1095             /* Analytical LJ-PME */
1096             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1097             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
1098             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
1099             exponent         = gmx_simd_exp_r(ewcljrsq);
1100             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1101             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1102             /* f6A = 6 * C6grid * (1 - poly) */
1103             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
1104             /* f6B = C6grid * exponent * beta^6 */
1105             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
1106             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1107             fvdw              = _mm_mul_ps(_mm_add_ps(_mm_mul_ps(_mm_sub_ps(_mm_mul_ps(c12_00,rinvsix),_mm_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1108
1109             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
1110
1111             fscal            = fvdw;
1112
1113             fscal            = _mm_and_ps(fscal,cutoff_mask);
1114
1115             /* Calculate temporary vectorial force */
1116             tx               = _mm_mul_ps(fscal,dx00);
1117             ty               = _mm_mul_ps(fscal,dy00);
1118             tz               = _mm_mul_ps(fscal,dz00);
1119
1120             /* Update vectorial force */
1121             fix0             = _mm_add_ps(fix0,tx);
1122             fiy0             = _mm_add_ps(fiy0,ty);
1123             fiz0             = _mm_add_ps(fiz0,tz);
1124
1125             fjx0             = _mm_add_ps(fjx0,tx);
1126             fjy0             = _mm_add_ps(fjy0,ty);
1127             fjz0             = _mm_add_ps(fjz0,tz);
1128
1129             }
1130
1131             /**************************
1132              * CALCULATE INTERACTIONS *
1133              **************************/
1134
1135             if (gmx_mm_any_lt(rsq10,rcutoff2))
1136             {
1137
1138             r10              = _mm_mul_ps(rsq10,rinv10);
1139
1140             /* Compute parameters for interactions between i and j atoms */
1141             qq10             = _mm_mul_ps(iq1,jq0);
1142
1143             /* EWALD ELECTROSTATICS */
1144
1145             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1146             ewrt             = _mm_mul_ps(r10,ewtabscale);
1147             ewitab           = _mm_cvttps_epi32(ewrt);
1148             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1149             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1150                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1151                                          &ewtabF,&ewtabFn);
1152             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1153             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1154
1155             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1156
1157             fscal            = felec;
1158
1159             fscal            = _mm_and_ps(fscal,cutoff_mask);
1160
1161             /* Calculate temporary vectorial force */
1162             tx               = _mm_mul_ps(fscal,dx10);
1163             ty               = _mm_mul_ps(fscal,dy10);
1164             tz               = _mm_mul_ps(fscal,dz10);
1165
1166             /* Update vectorial force */
1167             fix1             = _mm_add_ps(fix1,tx);
1168             fiy1             = _mm_add_ps(fiy1,ty);
1169             fiz1             = _mm_add_ps(fiz1,tz);
1170
1171             fjx0             = _mm_add_ps(fjx0,tx);
1172             fjy0             = _mm_add_ps(fjy0,ty);
1173             fjz0             = _mm_add_ps(fjz0,tz);
1174
1175             }
1176
1177             /**************************
1178              * CALCULATE INTERACTIONS *
1179              **************************/
1180
1181             if (gmx_mm_any_lt(rsq20,rcutoff2))
1182             {
1183
1184             r20              = _mm_mul_ps(rsq20,rinv20);
1185
1186             /* Compute parameters for interactions between i and j atoms */
1187             qq20             = _mm_mul_ps(iq2,jq0);
1188
1189             /* EWALD ELECTROSTATICS */
1190
1191             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1192             ewrt             = _mm_mul_ps(r20,ewtabscale);
1193             ewitab           = _mm_cvttps_epi32(ewrt);
1194             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1195             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1196                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1197                                          &ewtabF,&ewtabFn);
1198             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1199             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1200
1201             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1202
1203             fscal            = felec;
1204
1205             fscal            = _mm_and_ps(fscal,cutoff_mask);
1206
1207             /* Calculate temporary vectorial force */
1208             tx               = _mm_mul_ps(fscal,dx20);
1209             ty               = _mm_mul_ps(fscal,dy20);
1210             tz               = _mm_mul_ps(fscal,dz20);
1211
1212             /* Update vectorial force */
1213             fix2             = _mm_add_ps(fix2,tx);
1214             fiy2             = _mm_add_ps(fiy2,ty);
1215             fiz2             = _mm_add_ps(fiz2,tz);
1216
1217             fjx0             = _mm_add_ps(fjx0,tx);
1218             fjy0             = _mm_add_ps(fjy0,ty);
1219             fjz0             = _mm_add_ps(fjz0,tz);
1220
1221             }
1222
1223             /**************************
1224              * CALCULATE INTERACTIONS *
1225              **************************/
1226
1227             if (gmx_mm_any_lt(rsq30,rcutoff2))
1228             {
1229
1230             r30              = _mm_mul_ps(rsq30,rinv30);
1231
1232             /* Compute parameters for interactions between i and j atoms */
1233             qq30             = _mm_mul_ps(iq3,jq0);
1234
1235             /* EWALD ELECTROSTATICS */
1236
1237             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1238             ewrt             = _mm_mul_ps(r30,ewtabscale);
1239             ewitab           = _mm_cvttps_epi32(ewrt);
1240             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1241             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1242                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1243                                          &ewtabF,&ewtabFn);
1244             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1245             felec            = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
1246
1247             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
1248
1249             fscal            = felec;
1250
1251             fscal            = _mm_and_ps(fscal,cutoff_mask);
1252
1253             /* Calculate temporary vectorial force */
1254             tx               = _mm_mul_ps(fscal,dx30);
1255             ty               = _mm_mul_ps(fscal,dy30);
1256             tz               = _mm_mul_ps(fscal,dz30);
1257
1258             /* Update vectorial force */
1259             fix3             = _mm_add_ps(fix3,tx);
1260             fiy3             = _mm_add_ps(fiy3,ty);
1261             fiz3             = _mm_add_ps(fiz3,tz);
1262
1263             fjx0             = _mm_add_ps(fjx0,tx);
1264             fjy0             = _mm_add_ps(fjy0,ty);
1265             fjz0             = _mm_add_ps(fjz0,tz);
1266
1267             }
1268
1269             fjptrA             = f+j_coord_offsetA;
1270             fjptrB             = f+j_coord_offsetB;
1271             fjptrC             = f+j_coord_offsetC;
1272             fjptrD             = f+j_coord_offsetD;
1273
1274             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1275
1276             /* Inner loop uses 166 flops */
1277         }
1278
1279         if(jidx<j_index_end)
1280         {
1281
1282             /* Get j neighbor index, and coordinate index */
1283             jnrlistA         = jjnr[jidx];
1284             jnrlistB         = jjnr[jidx+1];
1285             jnrlistC         = jjnr[jidx+2];
1286             jnrlistD         = jjnr[jidx+3];
1287             /* Sign of each element will be negative for non-real atoms.
1288              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1289              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1290              */
1291             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1292             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1293             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1294             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1295             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1296             j_coord_offsetA  = DIM*jnrA;
1297             j_coord_offsetB  = DIM*jnrB;
1298             j_coord_offsetC  = DIM*jnrC;
1299             j_coord_offsetD  = DIM*jnrD;
1300
1301             /* load j atom coordinates */
1302             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1303                                               x+j_coord_offsetC,x+j_coord_offsetD,
1304                                               &jx0,&jy0,&jz0);
1305
1306             /* Calculate displacement vector */
1307             dx00             = _mm_sub_ps(ix0,jx0);
1308             dy00             = _mm_sub_ps(iy0,jy0);
1309             dz00             = _mm_sub_ps(iz0,jz0);
1310             dx10             = _mm_sub_ps(ix1,jx0);
1311             dy10             = _mm_sub_ps(iy1,jy0);
1312             dz10             = _mm_sub_ps(iz1,jz0);
1313             dx20             = _mm_sub_ps(ix2,jx0);
1314             dy20             = _mm_sub_ps(iy2,jy0);
1315             dz20             = _mm_sub_ps(iz2,jz0);
1316             dx30             = _mm_sub_ps(ix3,jx0);
1317             dy30             = _mm_sub_ps(iy3,jy0);
1318             dz30             = _mm_sub_ps(iz3,jz0);
1319
1320             /* Calculate squared distance and things based on it */
1321             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1322             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1323             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1324             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1325
1326             rinv00           = gmx_mm_invsqrt_ps(rsq00);
1327             rinv10           = gmx_mm_invsqrt_ps(rsq10);
1328             rinv20           = gmx_mm_invsqrt_ps(rsq20);
1329             rinv30           = gmx_mm_invsqrt_ps(rsq30);
1330
1331             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1332             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1333             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1334             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
1335
1336             /* Load parameters for j particles */
1337             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1338                                                               charge+jnrC+0,charge+jnrD+0);
1339             vdwjidx0A        = 2*vdwtype[jnrA+0];
1340             vdwjidx0B        = 2*vdwtype[jnrB+0];
1341             vdwjidx0C        = 2*vdwtype[jnrC+0];
1342             vdwjidx0D        = 2*vdwtype[jnrD+0];
1343
1344             fjx0             = _mm_setzero_ps();
1345             fjy0             = _mm_setzero_ps();
1346             fjz0             = _mm_setzero_ps();
1347
1348             /**************************
1349              * CALCULATE INTERACTIONS *
1350              **************************/
1351
1352             if (gmx_mm_any_lt(rsq00,rcutoff2))
1353             {
1354
1355             r00              = _mm_mul_ps(rsq00,rinv00);
1356             r00              = _mm_andnot_ps(dummy_mask,r00);
1357
1358             /* Compute parameters for interactions between i and j atoms */
1359             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1360                                          vdwparam+vdwioffset0+vdwjidx0B,
1361                                          vdwparam+vdwioffset0+vdwjidx0C,
1362                                          vdwparam+vdwioffset0+vdwjidx0D,
1363                                          &c6_00,&c12_00);
1364
1365             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
1366                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
1367                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
1368                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
1369
1370             /* Analytical LJ-PME */
1371             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1372             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
1373             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
1374             exponent         = gmx_simd_exp_r(ewcljrsq);
1375             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1376             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1377             /* f6A = 6 * C6grid * (1 - poly) */
1378             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
1379             /* f6B = C6grid * exponent * beta^6 */
1380             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
1381             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1382             fvdw              = _mm_mul_ps(_mm_add_ps(_mm_mul_ps(_mm_sub_ps(_mm_mul_ps(c12_00,rinvsix),_mm_sub_ps(c6_00,f6A)),rinvsix),f6B),rinvsq00);
1383
1384             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
1385
1386             fscal            = fvdw;
1387
1388             fscal            = _mm_and_ps(fscal,cutoff_mask);
1389
1390             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1391
1392             /* Calculate temporary vectorial force */
1393             tx               = _mm_mul_ps(fscal,dx00);
1394             ty               = _mm_mul_ps(fscal,dy00);
1395             tz               = _mm_mul_ps(fscal,dz00);
1396
1397             /* Update vectorial force */
1398             fix0             = _mm_add_ps(fix0,tx);
1399             fiy0             = _mm_add_ps(fiy0,ty);
1400             fiz0             = _mm_add_ps(fiz0,tz);
1401
1402             fjx0             = _mm_add_ps(fjx0,tx);
1403             fjy0             = _mm_add_ps(fjy0,ty);
1404             fjz0             = _mm_add_ps(fjz0,tz);
1405
1406             }
1407
1408             /**************************
1409              * CALCULATE INTERACTIONS *
1410              **************************/
1411
1412             if (gmx_mm_any_lt(rsq10,rcutoff2))
1413             {
1414
1415             r10              = _mm_mul_ps(rsq10,rinv10);
1416             r10              = _mm_andnot_ps(dummy_mask,r10);
1417
1418             /* Compute parameters for interactions between i and j atoms */
1419             qq10             = _mm_mul_ps(iq1,jq0);
1420
1421             /* EWALD ELECTROSTATICS */
1422
1423             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1424             ewrt             = _mm_mul_ps(r10,ewtabscale);
1425             ewitab           = _mm_cvttps_epi32(ewrt);
1426             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1427             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1428                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1429                                          &ewtabF,&ewtabFn);
1430             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1431             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1432
1433             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1434
1435             fscal            = felec;
1436
1437             fscal            = _mm_and_ps(fscal,cutoff_mask);
1438
1439             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1440
1441             /* Calculate temporary vectorial force */
1442             tx               = _mm_mul_ps(fscal,dx10);
1443             ty               = _mm_mul_ps(fscal,dy10);
1444             tz               = _mm_mul_ps(fscal,dz10);
1445
1446             /* Update vectorial force */
1447             fix1             = _mm_add_ps(fix1,tx);
1448             fiy1             = _mm_add_ps(fiy1,ty);
1449             fiz1             = _mm_add_ps(fiz1,tz);
1450
1451             fjx0             = _mm_add_ps(fjx0,tx);
1452             fjy0             = _mm_add_ps(fjy0,ty);
1453             fjz0             = _mm_add_ps(fjz0,tz);
1454
1455             }
1456
1457             /**************************
1458              * CALCULATE INTERACTIONS *
1459              **************************/
1460
1461             if (gmx_mm_any_lt(rsq20,rcutoff2))
1462             {
1463
1464             r20              = _mm_mul_ps(rsq20,rinv20);
1465             r20              = _mm_andnot_ps(dummy_mask,r20);
1466
1467             /* Compute parameters for interactions between i and j atoms */
1468             qq20             = _mm_mul_ps(iq2,jq0);
1469
1470             /* EWALD ELECTROSTATICS */
1471
1472             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1473             ewrt             = _mm_mul_ps(r20,ewtabscale);
1474             ewitab           = _mm_cvttps_epi32(ewrt);
1475             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1476             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1477                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1478                                          &ewtabF,&ewtabFn);
1479             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1480             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1481
1482             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1483
1484             fscal            = felec;
1485
1486             fscal            = _mm_and_ps(fscal,cutoff_mask);
1487
1488             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1489
1490             /* Calculate temporary vectorial force */
1491             tx               = _mm_mul_ps(fscal,dx20);
1492             ty               = _mm_mul_ps(fscal,dy20);
1493             tz               = _mm_mul_ps(fscal,dz20);
1494
1495             /* Update vectorial force */
1496             fix2             = _mm_add_ps(fix2,tx);
1497             fiy2             = _mm_add_ps(fiy2,ty);
1498             fiz2             = _mm_add_ps(fiz2,tz);
1499
1500             fjx0             = _mm_add_ps(fjx0,tx);
1501             fjy0             = _mm_add_ps(fjy0,ty);
1502             fjz0             = _mm_add_ps(fjz0,tz);
1503
1504             }
1505
1506             /**************************
1507              * CALCULATE INTERACTIONS *
1508              **************************/
1509
1510             if (gmx_mm_any_lt(rsq30,rcutoff2))
1511             {
1512
1513             r30              = _mm_mul_ps(rsq30,rinv30);
1514             r30              = _mm_andnot_ps(dummy_mask,r30);
1515
1516             /* Compute parameters for interactions between i and j atoms */
1517             qq30             = _mm_mul_ps(iq3,jq0);
1518
1519             /* EWALD ELECTROSTATICS */
1520
1521             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1522             ewrt             = _mm_mul_ps(r30,ewtabscale);
1523             ewitab           = _mm_cvttps_epi32(ewrt);
1524             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1525             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1526                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1527                                          &ewtabF,&ewtabFn);
1528             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1529             felec            = _mm_mul_ps(_mm_mul_ps(qq30,rinv30),_mm_sub_ps(rinvsq30,felec));
1530
1531             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
1532
1533             fscal            = felec;
1534
1535             fscal            = _mm_and_ps(fscal,cutoff_mask);
1536
1537             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1538
1539             /* Calculate temporary vectorial force */
1540             tx               = _mm_mul_ps(fscal,dx30);
1541             ty               = _mm_mul_ps(fscal,dy30);
1542             tz               = _mm_mul_ps(fscal,dz30);
1543
1544             /* Update vectorial force */
1545             fix3             = _mm_add_ps(fix3,tx);
1546             fiy3             = _mm_add_ps(fiy3,ty);
1547             fiz3             = _mm_add_ps(fiz3,tz);
1548
1549             fjx0             = _mm_add_ps(fjx0,tx);
1550             fjy0             = _mm_add_ps(fjy0,ty);
1551             fjz0             = _mm_add_ps(fjz0,tz);
1552
1553             }
1554
1555             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1556             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1557             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1558             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1559
1560             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1561
1562             /* Inner loop uses 170 flops */
1563         }
1564
1565         /* End of innermost loop */
1566
1567         gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1568                                               f+i_coord_offset,fshift+i_shift_offset);
1569
1570         /* Increment number of inner iterations */
1571         inneriter                  += j_index_end - j_index_start;
1572
1573         /* Outer loop uses 24 flops */
1574     }
1575
1576     /* Increment number of outer iterations */
1577     outeriter        += nri;
1578
1579     /* Update outer/inner flops */
1580
1581     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*170);
1582 }