src/gromacs/gmxlib/nonbonded/nb_kernel_sse4_1_single/nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_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 "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "nrnb.h"
  46
  47 #include "gromacs/simd/math_x86_sse4_1_single.h"
  48 #include "kernelutil_x86_sse4_1_single.h"
  49
  50 /*
  51  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_VF_sse4_1_single
  52  * Electrostatics interaction: Ewald
  53  * VdW interaction:            LJEwald
  54  * Geometry:                   Water3-Water3
  55  * Calculate force/pot:        PotentialAndForce
  56  */
  57 void
  58 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_VF_sse4_1_single
  59                     (t_nblist                    * gmx_restrict       nlist,
  60                      rvec                        * gmx_restrict          xx,
  61                      rvec                        * gmx_restrict          ff,
  62                      t_forcerec                  * gmx_restrict          fr,
  63                      t_mdatoms                   * gmx_restrict     mdatoms,
  64                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  65                      t_nrnb                      * gmx_restrict        nrnb)
  66 {
  67     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  68      * just 0 for non-waters.
  69      * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
  70      * jnr indices corresponding to data put in the four positions in the SIMD register.
  71      */
  72     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  73     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  74     int              jnrA,jnrB,jnrC,jnrD;
  75     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  76     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  77     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  78     real             rcutoff_scalar;
  79     real             *shiftvec,*fshift,*x,*f;
  80     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  81     real             scratch[4*DIM];
  82     __m128           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  83     int              vdwioffset0;
  84     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  85     int              vdwioffset1;
  86     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  87     int              vdwioffset2;
  88     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  89     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  90     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  91     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
  92     __m128           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  93     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
  94     __m128           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  95     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  96     __m128           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
  97     __m128           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
  98     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  99     __m128           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 100     __m128           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 101     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 102     __m128           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 103     __m128           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 104     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 105     real             *charge;
 106     int              nvdwtype;
 107     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 108     int              *vdwtype;
 109     real             *vdwparam;
 110     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 111     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 112     __m128           c6grid_00;
 113     __m128           c6grid_01;
 114     __m128           c6grid_02;
 115     __m128           c6grid_10;
 116     __m128           c6grid_11;
 117     __m128           c6grid_12;
 118     __m128           c6grid_20;
 119     __m128           c6grid_21;
 120     __m128           c6grid_22;
 121     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 122     real             *vdwgridparam;
 123     __m128           one_half  = _mm_set1_ps(0.5);
 124     __m128           minus_one = _mm_set1_ps(-1.0);
 125     __m128i          ewitab;
 126     __m128           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 127     real             *ewtab;
 128     __m128           dummy_mask,cutoff_mask;
 129     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 130     __m128           one     = _mm_set1_ps(1.0);
 131     __m128           two     = _mm_set1_ps(2.0);
 132     x                = xx[0];
 133     f                = ff[0];
 134
 135     nri              = nlist->nri;
 136     iinr             = nlist->iinr;
 137     jindex           = nlist->jindex;
 138     jjnr             = nlist->jjnr;
 139     shiftidx         = nlist->shift;
 140     gid              = nlist->gid;
 141     shiftvec         = fr->shift_vec[0];
 142     fshift           = fr->fshift[0];
 143     facel            = _mm_set1_ps(fr->epsfac);
 144     charge           = mdatoms->chargeA;
 145     nvdwtype         = fr->ntype;
 146     vdwparam         = fr->nbfp;
 147     vdwtype          = mdatoms->typeA;
 148     vdwgridparam     = fr->ljpme_c6grid;
 149     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
 150     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
 151     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
 152
 153     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 154     ewtab            = fr->ic->tabq_coul_FDV0;
 155     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 156     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 157
 158     /* Setup water-specific parameters */
 159     inr              = nlist->iinr[0];
 160     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
 161     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 162     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 163     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 164
 165     jq0              = _mm_set1_ps(charge[inr+0]);
 166     jq1              = _mm_set1_ps(charge[inr+1]);
 167     jq2              = _mm_set1_ps(charge[inr+2]);
 168     vdwjidx0A        = 2*vdwtype[inr+0];
 169     qq00             = _mm_mul_ps(iq0,jq0);
 170     c6_00            = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
 171     c12_00           = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
 172     c6grid_00        = _mm_set1_ps(vdwgridparam[vdwioffset0+vdwjidx0A]);
 173     qq01             = _mm_mul_ps(iq0,jq1);
 174     qq02             = _mm_mul_ps(iq0,jq2);
 175     qq10             = _mm_mul_ps(iq1,jq0);
 176     qq11             = _mm_mul_ps(iq1,jq1);
 177     qq12             = _mm_mul_ps(iq1,jq2);
 178     qq20             = _mm_mul_ps(iq2,jq0);
 179     qq21             = _mm_mul_ps(iq2,jq1);
 180     qq22             = _mm_mul_ps(iq2,jq2);
 181
 182     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 183     rcutoff_scalar   = fr->rcoulomb;
 184     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 185     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 186
 187     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
 188     rvdw             = _mm_set1_ps(fr->rvdw);
 189
 190     /* Avoid stupid compiler warnings */
 191     jnrA = jnrB = jnrC = jnrD = 0;
 192     j_coord_offsetA = 0;
 193     j_coord_offsetB = 0;
 194     j_coord_offsetC = 0;
 195     j_coord_offsetD = 0;
 196
 197     outeriter        = 0;
 198     inneriter        = 0;
 199
 200     for(iidx=0;iidx<4*DIM;iidx++)
 201     {
 202         scratch[iidx] = 0.0;
 203     }
 204
 205     /* Start outer loop over neighborlists */
 206     for(iidx=0; iidx<nri; iidx++)
 207     {
 208         /* Load shift vector for this list */
 209         i_shift_offset   = DIM*shiftidx[iidx];
 210
 211         /* Load limits for loop over neighbors */
 212         j_index_start    = jindex[iidx];
 213         j_index_end      = jindex[iidx+1];
 214
 215         /* Get outer coordinate index */
 216         inr              = iinr[iidx];
 217         i_coord_offset   = DIM*inr;
 218
 219         /* Load i particle coords and add shift vector */
 220         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 221                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 222
 223         fix0             = _mm_setzero_ps();
 224         fiy0             = _mm_setzero_ps();
 225         fiz0             = _mm_setzero_ps();
 226         fix1             = _mm_setzero_ps();
 227         fiy1             = _mm_setzero_ps();
 228         fiz1             = _mm_setzero_ps();
 229         fix2             = _mm_setzero_ps();
 230         fiy2             = _mm_setzero_ps();
 231         fiz2             = _mm_setzero_ps();
 232
 233         /* Reset potential sums */
 234         velecsum         = _mm_setzero_ps();
 235         vvdwsum          = _mm_setzero_ps();
 236
 237         /* Start inner kernel loop */
 238         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 239         {
 240
 241             /* Get j neighbor index, and coordinate index */
 242             jnrA             = jjnr[jidx];
 243             jnrB             = jjnr[jidx+1];
 244             jnrC             = jjnr[jidx+2];
 245             jnrD             = jjnr[jidx+3];
 246             j_coord_offsetA  = DIM*jnrA;
 247             j_coord_offsetB  = DIM*jnrB;
 248             j_coord_offsetC  = DIM*jnrC;
 249             j_coord_offsetD  = DIM*jnrD;
 250
 251             /* load j atom coordinates */
 252             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 253                                               x+j_coord_offsetC,x+j_coord_offsetD,
 254                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 255
 256             /* Calculate displacement vector */
 257             dx00             = _mm_sub_ps(ix0,jx0);
 258             dy00             = _mm_sub_ps(iy0,jy0);
 259             dz00             = _mm_sub_ps(iz0,jz0);
 260             dx01             = _mm_sub_ps(ix0,jx1);
 261             dy01             = _mm_sub_ps(iy0,jy1);
 262             dz01             = _mm_sub_ps(iz0,jz1);
 263             dx02             = _mm_sub_ps(ix0,jx2);
 264             dy02             = _mm_sub_ps(iy0,jy2);
 265             dz02             = _mm_sub_ps(iz0,jz2);
 266             dx10             = _mm_sub_ps(ix1,jx0);
 267             dy10             = _mm_sub_ps(iy1,jy0);
 268             dz10             = _mm_sub_ps(iz1,jz0);
 269             dx11             = _mm_sub_ps(ix1,jx1);
 270             dy11             = _mm_sub_ps(iy1,jy1);
 271             dz11             = _mm_sub_ps(iz1,jz1);
 272             dx12             = _mm_sub_ps(ix1,jx2);
 273             dy12             = _mm_sub_ps(iy1,jy2);
 274             dz12             = _mm_sub_ps(iz1,jz2);
 275             dx20             = _mm_sub_ps(ix2,jx0);
 276             dy20             = _mm_sub_ps(iy2,jy0);
 277             dz20             = _mm_sub_ps(iz2,jz0);
 278             dx21             = _mm_sub_ps(ix2,jx1);
 279             dy21             = _mm_sub_ps(iy2,jy1);
 280             dz21             = _mm_sub_ps(iz2,jz1);
 281             dx22             = _mm_sub_ps(ix2,jx2);
 282             dy22             = _mm_sub_ps(iy2,jy2);
 283             dz22             = _mm_sub_ps(iz2,jz2);
 284
 285             /* Calculate squared distance and things based on it */
 286             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 287             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
 288             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
 289             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 290             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
 291             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
 292             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 293             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
 294             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
 295
 296             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 297             rinv01           = gmx_mm_invsqrt_ps(rsq01);
 298             rinv02           = gmx_mm_invsqrt_ps(rsq02);
 299             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 300             rinv11           = gmx_mm_invsqrt_ps(rsq11);
 301             rinv12           = gmx_mm_invsqrt_ps(rsq12);
 302             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 303             rinv21           = gmx_mm_invsqrt_ps(rsq21);
 304             rinv22           = gmx_mm_invsqrt_ps(rsq22);
 305
 306             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 307             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
 308             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
 309             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 310             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
 311             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
 312             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 313             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
 314             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
 315
 316             fjx0             = _mm_setzero_ps();
 317             fjy0             = _mm_setzero_ps();
 318             fjz0             = _mm_setzero_ps();
 319             fjx1             = _mm_setzero_ps();
 320             fjy1             = _mm_setzero_ps();
 321             fjz1             = _mm_setzero_ps();
 322             fjx2             = _mm_setzero_ps();
 323             fjy2             = _mm_setzero_ps();
 324             fjz2             = _mm_setzero_ps();
 325
 326             /**************************
 327              * CALCULATE INTERACTIONS *
 328              **************************/
 329
 330             if (gmx_mm_any_lt(rsq00,rcutoff2))
 331             {
 332
 333             r00              = _mm_mul_ps(rsq00,rinv00);
 334
 335             /* EWALD ELECTROSTATICS */
 336
 337             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 338             ewrt             = _mm_mul_ps(r00,ewtabscale);
 339             ewitab           = _mm_cvttps_epi32(ewrt);
 340             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 341             ewitab           = _mm_slli_epi32(ewitab,2);
 342             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 343             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 344             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 345             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 346             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 347             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 348             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 349             velec            = _mm_mul_ps(qq00,_mm_sub_ps(_mm_sub_ps(rinv00,sh_ewald),velec));
 350             felec            = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
 351
 352             /* Analytical LJ-PME */
 353             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 354             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 355             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 356             exponent         = gmx_simd_exp_r(ewcljrsq);
 357             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 358             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 359             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 360             vvdw6            = _mm_mul_ps(_mm_sub_ps(c6_00,_mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly))),rinvsix);
 361             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 362             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),
 363                                _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));
 364             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 365             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);
 366
 367             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 368
 369             /* Update potential sum for this i atom from the interaction with this j atom. */
 370             velec            = _mm_and_ps(velec,cutoff_mask);
 371             velecsum         = _mm_add_ps(velecsum,velec);
 372             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 373             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 374
 375             fscal            = _mm_add_ps(felec,fvdw);
 376
 377             fscal            = _mm_and_ps(fscal,cutoff_mask);
 378
 379             /* Calculate temporary vectorial force */
 380             tx               = _mm_mul_ps(fscal,dx00);
 381             ty               = _mm_mul_ps(fscal,dy00);
 382             tz               = _mm_mul_ps(fscal,dz00);
 383
 384             /* Update vectorial force */
 385             fix0             = _mm_add_ps(fix0,tx);
 386             fiy0             = _mm_add_ps(fiy0,ty);
 387             fiz0             = _mm_add_ps(fiz0,tz);
 388
 389             fjx0             = _mm_add_ps(fjx0,tx);
 390             fjy0             = _mm_add_ps(fjy0,ty);
 391             fjz0             = _mm_add_ps(fjz0,tz);
 392
 393             }
 394
 395             /**************************
 396              * CALCULATE INTERACTIONS *
 397              **************************/
 398
 399             if (gmx_mm_any_lt(rsq01,rcutoff2))
 400             {
 401
 402             r01              = _mm_mul_ps(rsq01,rinv01);
 403
 404             /* EWALD ELECTROSTATICS */
 405
 406             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 407             ewrt             = _mm_mul_ps(r01,ewtabscale);
 408             ewitab           = _mm_cvttps_epi32(ewrt);
 409             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 410             ewitab           = _mm_slli_epi32(ewitab,2);
 411             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 412             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 413             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 414             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 415             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 416             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 417             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 418             velec            = _mm_mul_ps(qq01,_mm_sub_ps(_mm_sub_ps(rinv01,sh_ewald),velec));
 419             felec            = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec));
 420
 421             cutoff_mask      = _mm_cmplt_ps(rsq01,rcutoff2);
 422
 423             /* Update potential sum for this i atom from the interaction with this j atom. */
 424             velec            = _mm_and_ps(velec,cutoff_mask);
 425             velecsum         = _mm_add_ps(velecsum,velec);
 426
 427             fscal            = felec;
 428
 429             fscal            = _mm_and_ps(fscal,cutoff_mask);
 430
 431             /* Calculate temporary vectorial force */
 432             tx               = _mm_mul_ps(fscal,dx01);
 433             ty               = _mm_mul_ps(fscal,dy01);
 434             tz               = _mm_mul_ps(fscal,dz01);
 435
 436             /* Update vectorial force */
 437             fix0             = _mm_add_ps(fix0,tx);
 438             fiy0             = _mm_add_ps(fiy0,ty);
 439             fiz0             = _mm_add_ps(fiz0,tz);
 440
 441             fjx1             = _mm_add_ps(fjx1,tx);
 442             fjy1             = _mm_add_ps(fjy1,ty);
 443             fjz1             = _mm_add_ps(fjz1,tz);
 444
 445             }
 446
 447             /**************************
 448              * CALCULATE INTERACTIONS *
 449              **************************/
 450
 451             if (gmx_mm_any_lt(rsq02,rcutoff2))
 452             {
 453
 454             r02              = _mm_mul_ps(rsq02,rinv02);
 455
 456             /* EWALD ELECTROSTATICS */
 457
 458             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 459             ewrt             = _mm_mul_ps(r02,ewtabscale);
 460             ewitab           = _mm_cvttps_epi32(ewrt);
 461             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 462             ewitab           = _mm_slli_epi32(ewitab,2);
 463             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 464             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 465             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 466             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 467             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 468             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 469             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 470             velec            = _mm_mul_ps(qq02,_mm_sub_ps(_mm_sub_ps(rinv02,sh_ewald),velec));
 471             felec            = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec));
 472
 473             cutoff_mask      = _mm_cmplt_ps(rsq02,rcutoff2);
 474
 475             /* Update potential sum for this i atom from the interaction with this j atom. */
 476             velec            = _mm_and_ps(velec,cutoff_mask);
 477             velecsum         = _mm_add_ps(velecsum,velec);
 478
 479             fscal            = felec;
 480
 481             fscal            = _mm_and_ps(fscal,cutoff_mask);
 482
 483             /* Calculate temporary vectorial force */
 484             tx               = _mm_mul_ps(fscal,dx02);
 485             ty               = _mm_mul_ps(fscal,dy02);
 486             tz               = _mm_mul_ps(fscal,dz02);
 487
 488             /* Update vectorial force */
 489             fix0             = _mm_add_ps(fix0,tx);
 490             fiy0             = _mm_add_ps(fiy0,ty);
 491             fiz0             = _mm_add_ps(fiz0,tz);
 492
 493             fjx2             = _mm_add_ps(fjx2,tx);
 494             fjy2             = _mm_add_ps(fjy2,ty);
 495             fjz2             = _mm_add_ps(fjz2,tz);
 496
 497             }
 498
 499             /**************************
 500              * CALCULATE INTERACTIONS *
 501              **************************/
 502
 503             if (gmx_mm_any_lt(rsq10,rcutoff2))
 504             {
 505
 506             r10              = _mm_mul_ps(rsq10,rinv10);
 507
 508             /* EWALD ELECTROSTATICS */
 509
 510             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 511             ewrt             = _mm_mul_ps(r10,ewtabscale);
 512             ewitab           = _mm_cvttps_epi32(ewrt);
 513             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 514             ewitab           = _mm_slli_epi32(ewitab,2);
 515             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 516             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 517             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 518             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 519             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 520             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 521             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 522             velec            = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec));
 523             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
 524
 525             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 526
 527             /* Update potential sum for this i atom from the interaction with this j atom. */
 528             velec            = _mm_and_ps(velec,cutoff_mask);
 529             velecsum         = _mm_add_ps(velecsum,velec);
 530
 531             fscal            = felec;
 532
 533             fscal            = _mm_and_ps(fscal,cutoff_mask);
 534
 535             /* Calculate temporary vectorial force */
 536             tx               = _mm_mul_ps(fscal,dx10);
 537             ty               = _mm_mul_ps(fscal,dy10);
 538             tz               = _mm_mul_ps(fscal,dz10);
 539
 540             /* Update vectorial force */
 541             fix1             = _mm_add_ps(fix1,tx);
 542             fiy1             = _mm_add_ps(fiy1,ty);
 543             fiz1             = _mm_add_ps(fiz1,tz);
 544
 545             fjx0             = _mm_add_ps(fjx0,tx);
 546             fjy0             = _mm_add_ps(fjy0,ty);
 547             fjz0             = _mm_add_ps(fjz0,tz);
 548
 549             }
 550
 551             /**************************
 552              * CALCULATE INTERACTIONS *
 553              **************************/
 554
 555             if (gmx_mm_any_lt(rsq11,rcutoff2))
 556             {
 557
 558             r11              = _mm_mul_ps(rsq11,rinv11);
 559
 560             /* EWALD ELECTROSTATICS */
 561
 562             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 563             ewrt             = _mm_mul_ps(r11,ewtabscale);
 564             ewitab           = _mm_cvttps_epi32(ewrt);
 565             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 566             ewitab           = _mm_slli_epi32(ewitab,2);
 567             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 568             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 569             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 570             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 571             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 572             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 573             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 574             velec            = _mm_mul_ps(qq11,_mm_sub_ps(_mm_sub_ps(rinv11,sh_ewald),velec));
 575             felec            = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
 576
 577             cutoff_mask      = _mm_cmplt_ps(rsq11,rcutoff2);
 578
 579             /* Update potential sum for this i atom from the interaction with this j atom. */
 580             velec            = _mm_and_ps(velec,cutoff_mask);
 581             velecsum         = _mm_add_ps(velecsum,velec);
 582
 583             fscal            = felec;
 584
 585             fscal            = _mm_and_ps(fscal,cutoff_mask);
 586
 587             /* Calculate temporary vectorial force */
 588             tx               = _mm_mul_ps(fscal,dx11);
 589             ty               = _mm_mul_ps(fscal,dy11);
 590             tz               = _mm_mul_ps(fscal,dz11);
 591
 592             /* Update vectorial force */
 593             fix1             = _mm_add_ps(fix1,tx);
 594             fiy1             = _mm_add_ps(fiy1,ty);
 595             fiz1             = _mm_add_ps(fiz1,tz);
 596
 597             fjx1             = _mm_add_ps(fjx1,tx);
 598             fjy1             = _mm_add_ps(fjy1,ty);
 599             fjz1             = _mm_add_ps(fjz1,tz);
 600
 601             }
 602
 603             /**************************
 604              * CALCULATE INTERACTIONS *
 605              **************************/
 606
 607             if (gmx_mm_any_lt(rsq12,rcutoff2))
 608             {
 609
 610             r12              = _mm_mul_ps(rsq12,rinv12);
 611
 612             /* EWALD ELECTROSTATICS */
 613
 614             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 615             ewrt             = _mm_mul_ps(r12,ewtabscale);
 616             ewitab           = _mm_cvttps_epi32(ewrt);
 617             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 618             ewitab           = _mm_slli_epi32(ewitab,2);
 619             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 620             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 621             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 622             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 623             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 624             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 625             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 626             velec            = _mm_mul_ps(qq12,_mm_sub_ps(_mm_sub_ps(rinv12,sh_ewald),velec));
 627             felec            = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
 628
 629             cutoff_mask      = _mm_cmplt_ps(rsq12,rcutoff2);
 630
 631             /* Update potential sum for this i atom from the interaction with this j atom. */
 632             velec            = _mm_and_ps(velec,cutoff_mask);
 633             velecsum         = _mm_add_ps(velecsum,velec);
 634
 635             fscal            = felec;
 636
 637             fscal            = _mm_and_ps(fscal,cutoff_mask);
 638
 639             /* Calculate temporary vectorial force */
 640             tx               = _mm_mul_ps(fscal,dx12);
 641             ty               = _mm_mul_ps(fscal,dy12);
 642             tz               = _mm_mul_ps(fscal,dz12);
 643
 644             /* Update vectorial force */
 645             fix1             = _mm_add_ps(fix1,tx);
 646             fiy1             = _mm_add_ps(fiy1,ty);
 647             fiz1             = _mm_add_ps(fiz1,tz);
 648
 649             fjx2             = _mm_add_ps(fjx2,tx);
 650             fjy2             = _mm_add_ps(fjy2,ty);
 651             fjz2             = _mm_add_ps(fjz2,tz);
 652
 653             }
 654
 655             /**************************
 656              * CALCULATE INTERACTIONS *
 657              **************************/
 658
 659             if (gmx_mm_any_lt(rsq20,rcutoff2))
 660             {
 661
 662             r20              = _mm_mul_ps(rsq20,rinv20);
 663
 664             /* EWALD ELECTROSTATICS */
 665
 666             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 667             ewrt             = _mm_mul_ps(r20,ewtabscale);
 668             ewitab           = _mm_cvttps_epi32(ewrt);
 669             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 670             ewitab           = _mm_slli_epi32(ewitab,2);
 671             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 672             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 673             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 674             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 675             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 676             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 677             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 678             velec            = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec));
 679             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
 680
 681             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 682
 683             /* Update potential sum for this i atom from the interaction with this j atom. */
 684             velec            = _mm_and_ps(velec,cutoff_mask);
 685             velecsum         = _mm_add_ps(velecsum,velec);
 686
 687             fscal            = felec;
 688
 689             fscal            = _mm_and_ps(fscal,cutoff_mask);
 690
 691             /* Calculate temporary vectorial force */
 692             tx               = _mm_mul_ps(fscal,dx20);
 693             ty               = _mm_mul_ps(fscal,dy20);
 694             tz               = _mm_mul_ps(fscal,dz20);
 695
 696             /* Update vectorial force */
 697             fix2             = _mm_add_ps(fix2,tx);
 698             fiy2             = _mm_add_ps(fiy2,ty);
 699             fiz2             = _mm_add_ps(fiz2,tz);
 700
 701             fjx0             = _mm_add_ps(fjx0,tx);
 702             fjy0             = _mm_add_ps(fjy0,ty);
 703             fjz0             = _mm_add_ps(fjz0,tz);
 704
 705             }
 706
 707             /**************************
 708              * CALCULATE INTERACTIONS *
 709              **************************/
 710
 711             if (gmx_mm_any_lt(rsq21,rcutoff2))
 712             {
 713
 714             r21              = _mm_mul_ps(rsq21,rinv21);
 715
 716             /* EWALD ELECTROSTATICS */
 717
 718             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 719             ewrt             = _mm_mul_ps(r21,ewtabscale);
 720             ewitab           = _mm_cvttps_epi32(ewrt);
 721             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 722             ewitab           = _mm_slli_epi32(ewitab,2);
 723             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 724             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 725             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 726             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 727             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 728             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 729             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 730             velec            = _mm_mul_ps(qq21,_mm_sub_ps(_mm_sub_ps(rinv21,sh_ewald),velec));
 731             felec            = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
 732
 733             cutoff_mask      = _mm_cmplt_ps(rsq21,rcutoff2);
 734
 735             /* Update potential sum for this i atom from the interaction with this j atom. */
 736             velec            = _mm_and_ps(velec,cutoff_mask);
 737             velecsum         = _mm_add_ps(velecsum,velec);
 738
 739             fscal            = felec;
 740
 741             fscal            = _mm_and_ps(fscal,cutoff_mask);
 742
 743             /* Calculate temporary vectorial force */
 744             tx               = _mm_mul_ps(fscal,dx21);
 745             ty               = _mm_mul_ps(fscal,dy21);
 746             tz               = _mm_mul_ps(fscal,dz21);
 747
 748             /* Update vectorial force */
 749             fix2             = _mm_add_ps(fix2,tx);
 750             fiy2             = _mm_add_ps(fiy2,ty);
 751             fiz2             = _mm_add_ps(fiz2,tz);
 752
 753             fjx1             = _mm_add_ps(fjx1,tx);
 754             fjy1             = _mm_add_ps(fjy1,ty);
 755             fjz1             = _mm_add_ps(fjz1,tz);
 756
 757             }
 758
 759             /**************************
 760              * CALCULATE INTERACTIONS *
 761              **************************/
 762
 763             if (gmx_mm_any_lt(rsq22,rcutoff2))
 764             {
 765
 766             r22              = _mm_mul_ps(rsq22,rinv22);
 767
 768             /* EWALD ELECTROSTATICS */
 769
 770             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 771             ewrt             = _mm_mul_ps(r22,ewtabscale);
 772             ewitab           = _mm_cvttps_epi32(ewrt);
 773             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 774             ewitab           = _mm_slli_epi32(ewitab,2);
 775             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 776             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 777             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 778             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 779             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 780             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 781             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 782             velec            = _mm_mul_ps(qq22,_mm_sub_ps(_mm_sub_ps(rinv22,sh_ewald),velec));
 783             felec            = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
 784
 785             cutoff_mask      = _mm_cmplt_ps(rsq22,rcutoff2);
 786
 787             /* Update potential sum for this i atom from the interaction with this j atom. */
 788             velec            = _mm_and_ps(velec,cutoff_mask);
 789             velecsum         = _mm_add_ps(velecsum,velec);
 790
 791             fscal            = felec;
 792
 793             fscal            = _mm_and_ps(fscal,cutoff_mask);
 794
 795             /* Calculate temporary vectorial force */
 796             tx               = _mm_mul_ps(fscal,dx22);
 797             ty               = _mm_mul_ps(fscal,dy22);
 798             tz               = _mm_mul_ps(fscal,dz22);
 799
 800             /* Update vectorial force */
 801             fix2             = _mm_add_ps(fix2,tx);
 802             fiy2             = _mm_add_ps(fiy2,ty);
 803             fiz2             = _mm_add_ps(fiz2,tz);
 804
 805             fjx2             = _mm_add_ps(fjx2,tx);
 806             fjy2             = _mm_add_ps(fjy2,ty);
 807             fjz2             = _mm_add_ps(fjz2,tz);
 808
 809             }
 810
 811             fjptrA             = f+j_coord_offsetA;
 812             fjptrB             = f+j_coord_offsetB;
 813             fjptrC             = f+j_coord_offsetC;
 814             fjptrD             = f+j_coord_offsetD;
 815
 816             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 817                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 818
 819             /* Inner loop uses 450 flops */
 820         }
 821
 822         if(jidx<j_index_end)
 823         {
 824
 825             /* Get j neighbor index, and coordinate index */
 826             jnrlistA         = jjnr[jidx];
 827             jnrlistB         = jjnr[jidx+1];
 828             jnrlistC         = jjnr[jidx+2];
 829             jnrlistD         = jjnr[jidx+3];
 830             /* Sign of each element will be negative for non-real atoms.
 831              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 832              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 833              */
 834             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 835             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 836             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 837             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 838             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 839             j_coord_offsetA  = DIM*jnrA;
 840             j_coord_offsetB  = DIM*jnrB;
 841             j_coord_offsetC  = DIM*jnrC;
 842             j_coord_offsetD  = DIM*jnrD;
 843
 844             /* load j atom coordinates */
 845             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 846                                               x+j_coord_offsetC,x+j_coord_offsetD,
 847                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 848
 849             /* Calculate displacement vector */
 850             dx00             = _mm_sub_ps(ix0,jx0);
 851             dy00             = _mm_sub_ps(iy0,jy0);
 852             dz00             = _mm_sub_ps(iz0,jz0);
 853             dx01             = _mm_sub_ps(ix0,jx1);
 854             dy01             = _mm_sub_ps(iy0,jy1);
 855             dz01             = _mm_sub_ps(iz0,jz1);
 856             dx02             = _mm_sub_ps(ix0,jx2);
 857             dy02             = _mm_sub_ps(iy0,jy2);
 858             dz02             = _mm_sub_ps(iz0,jz2);
 859             dx10             = _mm_sub_ps(ix1,jx0);
 860             dy10             = _mm_sub_ps(iy1,jy0);
 861             dz10             = _mm_sub_ps(iz1,jz0);
 862             dx11             = _mm_sub_ps(ix1,jx1);
 863             dy11             = _mm_sub_ps(iy1,jy1);
 864             dz11             = _mm_sub_ps(iz1,jz1);
 865             dx12             = _mm_sub_ps(ix1,jx2);
 866             dy12             = _mm_sub_ps(iy1,jy2);
 867             dz12             = _mm_sub_ps(iz1,jz2);
 868             dx20             = _mm_sub_ps(ix2,jx0);
 869             dy20             = _mm_sub_ps(iy2,jy0);
 870             dz20             = _mm_sub_ps(iz2,jz0);
 871             dx21             = _mm_sub_ps(ix2,jx1);
 872             dy21             = _mm_sub_ps(iy2,jy1);
 873             dz21             = _mm_sub_ps(iz2,jz1);
 874             dx22             = _mm_sub_ps(ix2,jx2);
 875             dy22             = _mm_sub_ps(iy2,jy2);
 876             dz22             = _mm_sub_ps(iz2,jz2);
 877
 878             /* Calculate squared distance and things based on it */
 879             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 880             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
 881             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
 882             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 883             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
 884             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
 885             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 886             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
 887             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
 888
 889             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 890             rinv01           = gmx_mm_invsqrt_ps(rsq01);
 891             rinv02           = gmx_mm_invsqrt_ps(rsq02);
 892             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 893             rinv11           = gmx_mm_invsqrt_ps(rsq11);
 894             rinv12           = gmx_mm_invsqrt_ps(rsq12);
 895             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 896             rinv21           = gmx_mm_invsqrt_ps(rsq21);
 897             rinv22           = gmx_mm_invsqrt_ps(rsq22);
 898
 899             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 900             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
 901             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
 902             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 903             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
 904             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
 905             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 906             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
 907             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
 908
 909             fjx0             = _mm_setzero_ps();
 910             fjy0             = _mm_setzero_ps();
 911             fjz0             = _mm_setzero_ps();
 912             fjx1             = _mm_setzero_ps();
 913             fjy1             = _mm_setzero_ps();
 914             fjz1             = _mm_setzero_ps();
 915             fjx2             = _mm_setzero_ps();
 916             fjy2             = _mm_setzero_ps();
 917             fjz2             = _mm_setzero_ps();
 918
 919             /**************************
 920              * CALCULATE INTERACTIONS *
 921              **************************/
 922
 923             if (gmx_mm_any_lt(rsq00,rcutoff2))
 924             {
 925
 926             r00              = _mm_mul_ps(rsq00,rinv00);
 927             r00              = _mm_andnot_ps(dummy_mask,r00);
 928
 929             /* EWALD ELECTROSTATICS */
 930
 931             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 932             ewrt             = _mm_mul_ps(r00,ewtabscale);
 933             ewitab           = _mm_cvttps_epi32(ewrt);
 934             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
 935             ewitab           = _mm_slli_epi32(ewitab,2);
 936             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
 937             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
 938             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
 939             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
 940             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
 941             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
 942             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
 943             velec            = _mm_mul_ps(qq00,_mm_sub_ps(_mm_sub_ps(rinv00,sh_ewald),velec));
 944             felec            = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
 945
 946             /* Analytical LJ-PME */
 947             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 948             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 949             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 950             exponent         = gmx_simd_exp_r(ewcljrsq);
 951             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 952             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
 953             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 954             vvdw6            = _mm_mul_ps(_mm_sub_ps(c6_00,_mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly))),rinvsix);
 955             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 956             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),
 957                                _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));
 958             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 959             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);
 960
 961             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 962
 963             /* Update potential sum for this i atom from the interaction with this j atom. */
 964             velec            = _mm_and_ps(velec,cutoff_mask);
 965             velec            = _mm_andnot_ps(dummy_mask,velec);
 966             velecsum         = _mm_add_ps(velecsum,velec);
 967             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 968             vvdw             = _mm_andnot_ps(dummy_mask,vvdw);
 969             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 970
 971             fscal            = _mm_add_ps(felec,fvdw);
 972
 973             fscal            = _mm_and_ps(fscal,cutoff_mask);
 974
 975             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 976
 977             /* Calculate temporary vectorial force */
 978             tx               = _mm_mul_ps(fscal,dx00);
 979             ty               = _mm_mul_ps(fscal,dy00);
 980             tz               = _mm_mul_ps(fscal,dz00);
 981
 982             /* Update vectorial force */
 983             fix0             = _mm_add_ps(fix0,tx);
 984             fiy0             = _mm_add_ps(fiy0,ty);
 985             fiz0             = _mm_add_ps(fiz0,tz);
 986
 987             fjx0             = _mm_add_ps(fjx0,tx);
 988             fjy0             = _mm_add_ps(fjy0,ty);
 989             fjz0             = _mm_add_ps(fjz0,tz);
 990
 991             }
 992
 993             /**************************
 994              * CALCULATE INTERACTIONS *
 995              **************************/
 996
 997             if (gmx_mm_any_lt(rsq01,rcutoff2))
 998             {
 999
1000             r01              = _mm_mul_ps(rsq01,rinv01);
1001             r01              = _mm_andnot_ps(dummy_mask,r01);
1002
1003             /* EWALD ELECTROSTATICS */
1004
1005             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1006             ewrt             = _mm_mul_ps(r01,ewtabscale);
1007             ewitab           = _mm_cvttps_epi32(ewrt);
1008             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1009             ewitab           = _mm_slli_epi32(ewitab,2);
1010             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1011             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1012             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1013             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1014             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1015             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1016             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1017             velec            = _mm_mul_ps(qq01,_mm_sub_ps(_mm_sub_ps(rinv01,sh_ewald),velec));
1018             felec            = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec));
1019
1020             cutoff_mask      = _mm_cmplt_ps(rsq01,rcutoff2);
1021
1022             /* Update potential sum for this i atom from the interaction with this j atom. */
1023             velec            = _mm_and_ps(velec,cutoff_mask);
1024             velec            = _mm_andnot_ps(dummy_mask,velec);
1025             velecsum         = _mm_add_ps(velecsum,velec);
1026
1027             fscal            = felec;
1028
1029             fscal            = _mm_and_ps(fscal,cutoff_mask);
1030
1031             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1032
1033             /* Calculate temporary vectorial force */
1034             tx               = _mm_mul_ps(fscal,dx01);
1035             ty               = _mm_mul_ps(fscal,dy01);
1036             tz               = _mm_mul_ps(fscal,dz01);
1037
1038             /* Update vectorial force */
1039             fix0             = _mm_add_ps(fix0,tx);
1040             fiy0             = _mm_add_ps(fiy0,ty);
1041             fiz0             = _mm_add_ps(fiz0,tz);
1042
1043             fjx1             = _mm_add_ps(fjx1,tx);
1044             fjy1             = _mm_add_ps(fjy1,ty);
1045             fjz1             = _mm_add_ps(fjz1,tz);
1046
1047             }
1048
1049             /**************************
1050              * CALCULATE INTERACTIONS *
1051              **************************/
1052
1053             if (gmx_mm_any_lt(rsq02,rcutoff2))
1054             {
1055
1056             r02              = _mm_mul_ps(rsq02,rinv02);
1057             r02              = _mm_andnot_ps(dummy_mask,r02);
1058
1059             /* EWALD ELECTROSTATICS */
1060
1061             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1062             ewrt             = _mm_mul_ps(r02,ewtabscale);
1063             ewitab           = _mm_cvttps_epi32(ewrt);
1064             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1065             ewitab           = _mm_slli_epi32(ewitab,2);
1066             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1067             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1068             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1069             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1070             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1071             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1072             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1073             velec            = _mm_mul_ps(qq02,_mm_sub_ps(_mm_sub_ps(rinv02,sh_ewald),velec));
1074             felec            = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec));
1075
1076             cutoff_mask      = _mm_cmplt_ps(rsq02,rcutoff2);
1077
1078             /* Update potential sum for this i atom from the interaction with this j atom. */
1079             velec            = _mm_and_ps(velec,cutoff_mask);
1080             velec            = _mm_andnot_ps(dummy_mask,velec);
1081             velecsum         = _mm_add_ps(velecsum,velec);
1082
1083             fscal            = felec;
1084
1085             fscal            = _mm_and_ps(fscal,cutoff_mask);
1086
1087             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1088
1089             /* Calculate temporary vectorial force */
1090             tx               = _mm_mul_ps(fscal,dx02);
1091             ty               = _mm_mul_ps(fscal,dy02);
1092             tz               = _mm_mul_ps(fscal,dz02);
1093
1094             /* Update vectorial force */
1095             fix0             = _mm_add_ps(fix0,tx);
1096             fiy0             = _mm_add_ps(fiy0,ty);
1097             fiz0             = _mm_add_ps(fiz0,tz);
1098
1099             fjx2             = _mm_add_ps(fjx2,tx);
1100             fjy2             = _mm_add_ps(fjy2,ty);
1101             fjz2             = _mm_add_ps(fjz2,tz);
1102
1103             }
1104
1105             /**************************
1106              * CALCULATE INTERACTIONS *
1107              **************************/
1108
1109             if (gmx_mm_any_lt(rsq10,rcutoff2))
1110             {
1111
1112             r10              = _mm_mul_ps(rsq10,rinv10);
1113             r10              = _mm_andnot_ps(dummy_mask,r10);
1114
1115             /* EWALD ELECTROSTATICS */
1116
1117             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1118             ewrt             = _mm_mul_ps(r10,ewtabscale);
1119             ewitab           = _mm_cvttps_epi32(ewrt);
1120             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1121             ewitab           = _mm_slli_epi32(ewitab,2);
1122             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1123             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1124             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1125             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1126             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1127             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1128             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1129             velec            = _mm_mul_ps(qq10,_mm_sub_ps(_mm_sub_ps(rinv10,sh_ewald),velec));
1130             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1131
1132             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1133
1134             /* Update potential sum for this i atom from the interaction with this j atom. */
1135             velec            = _mm_and_ps(velec,cutoff_mask);
1136             velec            = _mm_andnot_ps(dummy_mask,velec);
1137             velecsum         = _mm_add_ps(velecsum,velec);
1138
1139             fscal            = felec;
1140
1141             fscal            = _mm_and_ps(fscal,cutoff_mask);
1142
1143             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1144
1145             /* Calculate temporary vectorial force */
1146             tx               = _mm_mul_ps(fscal,dx10);
1147             ty               = _mm_mul_ps(fscal,dy10);
1148             tz               = _mm_mul_ps(fscal,dz10);
1149
1150             /* Update vectorial force */
1151             fix1             = _mm_add_ps(fix1,tx);
1152             fiy1             = _mm_add_ps(fiy1,ty);
1153             fiz1             = _mm_add_ps(fiz1,tz);
1154
1155             fjx0             = _mm_add_ps(fjx0,tx);
1156             fjy0             = _mm_add_ps(fjy0,ty);
1157             fjz0             = _mm_add_ps(fjz0,tz);
1158
1159             }
1160
1161             /**************************
1162              * CALCULATE INTERACTIONS *
1163              **************************/
1164
1165             if (gmx_mm_any_lt(rsq11,rcutoff2))
1166             {
1167
1168             r11              = _mm_mul_ps(rsq11,rinv11);
1169             r11              = _mm_andnot_ps(dummy_mask,r11);
1170
1171             /* EWALD ELECTROSTATICS */
1172
1173             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1174             ewrt             = _mm_mul_ps(r11,ewtabscale);
1175             ewitab           = _mm_cvttps_epi32(ewrt);
1176             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1177             ewitab           = _mm_slli_epi32(ewitab,2);
1178             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1179             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1180             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1181             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1182             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1183             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1184             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1185             velec            = _mm_mul_ps(qq11,_mm_sub_ps(_mm_sub_ps(rinv11,sh_ewald),velec));
1186             felec            = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
1187
1188             cutoff_mask      = _mm_cmplt_ps(rsq11,rcutoff2);
1189
1190             /* Update potential sum for this i atom from the interaction with this j atom. */
1191             velec            = _mm_and_ps(velec,cutoff_mask);
1192             velec            = _mm_andnot_ps(dummy_mask,velec);
1193             velecsum         = _mm_add_ps(velecsum,velec);
1194
1195             fscal            = felec;
1196
1197             fscal            = _mm_and_ps(fscal,cutoff_mask);
1198
1199             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1200
1201             /* Calculate temporary vectorial force */
1202             tx               = _mm_mul_ps(fscal,dx11);
1203             ty               = _mm_mul_ps(fscal,dy11);
1204             tz               = _mm_mul_ps(fscal,dz11);
1205
1206             /* Update vectorial force */
1207             fix1             = _mm_add_ps(fix1,tx);
1208             fiy1             = _mm_add_ps(fiy1,ty);
1209             fiz1             = _mm_add_ps(fiz1,tz);
1210
1211             fjx1             = _mm_add_ps(fjx1,tx);
1212             fjy1             = _mm_add_ps(fjy1,ty);
1213             fjz1             = _mm_add_ps(fjz1,tz);
1214
1215             }
1216
1217             /**************************
1218              * CALCULATE INTERACTIONS *
1219              **************************/
1220
1221             if (gmx_mm_any_lt(rsq12,rcutoff2))
1222             {
1223
1224             r12              = _mm_mul_ps(rsq12,rinv12);
1225             r12              = _mm_andnot_ps(dummy_mask,r12);
1226
1227             /* EWALD ELECTROSTATICS */
1228
1229             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1230             ewrt             = _mm_mul_ps(r12,ewtabscale);
1231             ewitab           = _mm_cvttps_epi32(ewrt);
1232             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1233             ewitab           = _mm_slli_epi32(ewitab,2);
1234             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1235             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1236             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1237             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1238             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1239             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1240             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1241             velec            = _mm_mul_ps(qq12,_mm_sub_ps(_mm_sub_ps(rinv12,sh_ewald),velec));
1242             felec            = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
1243
1244             cutoff_mask      = _mm_cmplt_ps(rsq12,rcutoff2);
1245
1246             /* Update potential sum for this i atom from the interaction with this j atom. */
1247             velec            = _mm_and_ps(velec,cutoff_mask);
1248             velec            = _mm_andnot_ps(dummy_mask,velec);
1249             velecsum         = _mm_add_ps(velecsum,velec);
1250
1251             fscal            = felec;
1252
1253             fscal            = _mm_and_ps(fscal,cutoff_mask);
1254
1255             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1256
1257             /* Calculate temporary vectorial force */
1258             tx               = _mm_mul_ps(fscal,dx12);
1259             ty               = _mm_mul_ps(fscal,dy12);
1260             tz               = _mm_mul_ps(fscal,dz12);
1261
1262             /* Update vectorial force */
1263             fix1             = _mm_add_ps(fix1,tx);
1264             fiy1             = _mm_add_ps(fiy1,ty);
1265             fiz1             = _mm_add_ps(fiz1,tz);
1266
1267             fjx2             = _mm_add_ps(fjx2,tx);
1268             fjy2             = _mm_add_ps(fjy2,ty);
1269             fjz2             = _mm_add_ps(fjz2,tz);
1270
1271             }
1272
1273             /**************************
1274              * CALCULATE INTERACTIONS *
1275              **************************/
1276
1277             if (gmx_mm_any_lt(rsq20,rcutoff2))
1278             {
1279
1280             r20              = _mm_mul_ps(rsq20,rinv20);
1281             r20              = _mm_andnot_ps(dummy_mask,r20);
1282
1283             /* EWALD ELECTROSTATICS */
1284
1285             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1286             ewrt             = _mm_mul_ps(r20,ewtabscale);
1287             ewitab           = _mm_cvttps_epi32(ewrt);
1288             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1289             ewitab           = _mm_slli_epi32(ewitab,2);
1290             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1291             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1292             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1293             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1294             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1295             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1296             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1297             velec            = _mm_mul_ps(qq20,_mm_sub_ps(_mm_sub_ps(rinv20,sh_ewald),velec));
1298             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
1299
1300             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1301
1302             /* Update potential sum for this i atom from the interaction with this j atom. */
1303             velec            = _mm_and_ps(velec,cutoff_mask);
1304             velec            = _mm_andnot_ps(dummy_mask,velec);
1305             velecsum         = _mm_add_ps(velecsum,velec);
1306
1307             fscal            = felec;
1308
1309             fscal            = _mm_and_ps(fscal,cutoff_mask);
1310
1311             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1312
1313             /* Calculate temporary vectorial force */
1314             tx               = _mm_mul_ps(fscal,dx20);
1315             ty               = _mm_mul_ps(fscal,dy20);
1316             tz               = _mm_mul_ps(fscal,dz20);
1317
1318             /* Update vectorial force */
1319             fix2             = _mm_add_ps(fix2,tx);
1320             fiy2             = _mm_add_ps(fiy2,ty);
1321             fiz2             = _mm_add_ps(fiz2,tz);
1322
1323             fjx0             = _mm_add_ps(fjx0,tx);
1324             fjy0             = _mm_add_ps(fjy0,ty);
1325             fjz0             = _mm_add_ps(fjz0,tz);
1326
1327             }
1328
1329             /**************************
1330              * CALCULATE INTERACTIONS *
1331              **************************/
1332
1333             if (gmx_mm_any_lt(rsq21,rcutoff2))
1334             {
1335
1336             r21              = _mm_mul_ps(rsq21,rinv21);
1337             r21              = _mm_andnot_ps(dummy_mask,r21);
1338
1339             /* EWALD ELECTROSTATICS */
1340
1341             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1342             ewrt             = _mm_mul_ps(r21,ewtabscale);
1343             ewitab           = _mm_cvttps_epi32(ewrt);
1344             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1345             ewitab           = _mm_slli_epi32(ewitab,2);
1346             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1347             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1348             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1349             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1350             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1351             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1352             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1353             velec            = _mm_mul_ps(qq21,_mm_sub_ps(_mm_sub_ps(rinv21,sh_ewald),velec));
1354             felec            = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
1355
1356             cutoff_mask      = _mm_cmplt_ps(rsq21,rcutoff2);
1357
1358             /* Update potential sum for this i atom from the interaction with this j atom. */
1359             velec            = _mm_and_ps(velec,cutoff_mask);
1360             velec            = _mm_andnot_ps(dummy_mask,velec);
1361             velecsum         = _mm_add_ps(velecsum,velec);
1362
1363             fscal            = felec;
1364
1365             fscal            = _mm_and_ps(fscal,cutoff_mask);
1366
1367             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1368
1369             /* Calculate temporary vectorial force */
1370             tx               = _mm_mul_ps(fscal,dx21);
1371             ty               = _mm_mul_ps(fscal,dy21);
1372             tz               = _mm_mul_ps(fscal,dz21);
1373
1374             /* Update vectorial force */
1375             fix2             = _mm_add_ps(fix2,tx);
1376             fiy2             = _mm_add_ps(fiy2,ty);
1377             fiz2             = _mm_add_ps(fiz2,tz);
1378
1379             fjx1             = _mm_add_ps(fjx1,tx);
1380             fjy1             = _mm_add_ps(fjy1,ty);
1381             fjz1             = _mm_add_ps(fjz1,tz);
1382
1383             }
1384
1385             /**************************
1386              * CALCULATE INTERACTIONS *
1387              **************************/
1388
1389             if (gmx_mm_any_lt(rsq22,rcutoff2))
1390             {
1391
1392             r22              = _mm_mul_ps(rsq22,rinv22);
1393             r22              = _mm_andnot_ps(dummy_mask,r22);
1394
1395             /* EWALD ELECTROSTATICS */
1396
1397             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1398             ewrt             = _mm_mul_ps(r22,ewtabscale);
1399             ewitab           = _mm_cvttps_epi32(ewrt);
1400             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1401             ewitab           = _mm_slli_epi32(ewitab,2);
1402             ewtabF           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,0) );
1403             ewtabD           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,1) );
1404             ewtabV           = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,2) );
1405             ewtabFn          = _mm_load_ps( ewtab + gmx_mm_extract_epi32(ewitab,3) );
1406             _MM_TRANSPOSE4_PS(ewtabF,ewtabD,ewtabV,ewtabFn);
1407             felec            = _mm_add_ps(ewtabF,_mm_mul_ps(eweps,ewtabD));
1408             velec            = _mm_sub_ps(ewtabV,_mm_mul_ps(_mm_mul_ps(ewtabhalfspace,eweps),_mm_add_ps(ewtabF,felec)));
1409             velec            = _mm_mul_ps(qq22,_mm_sub_ps(_mm_sub_ps(rinv22,sh_ewald),velec));
1410             felec            = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
1411
1412             cutoff_mask      = _mm_cmplt_ps(rsq22,rcutoff2);
1413
1414             /* Update potential sum for this i atom from the interaction with this j atom. */
1415             velec            = _mm_and_ps(velec,cutoff_mask);
1416             velec            = _mm_andnot_ps(dummy_mask,velec);
1417             velecsum         = _mm_add_ps(velecsum,velec);
1418
1419             fscal            = felec;
1420
1421             fscal            = _mm_and_ps(fscal,cutoff_mask);
1422
1423             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1424
1425             /* Calculate temporary vectorial force */
1426             tx               = _mm_mul_ps(fscal,dx22);
1427             ty               = _mm_mul_ps(fscal,dy22);
1428             tz               = _mm_mul_ps(fscal,dz22);
1429
1430             /* Update vectorial force */
1431             fix2             = _mm_add_ps(fix2,tx);
1432             fiy2             = _mm_add_ps(fiy2,ty);
1433             fiz2             = _mm_add_ps(fiz2,tz);
1434
1435             fjx2             = _mm_add_ps(fjx2,tx);
1436             fjy2             = _mm_add_ps(fjy2,ty);
1437             fjz2             = _mm_add_ps(fjz2,tz);
1438
1439             }
1440
1441             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1442             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1443             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1444             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1445
1446             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1447                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1448
1449             /* Inner loop uses 459 flops */
1450         }
1451
1452         /* End of innermost loop */
1453
1454         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1455                                               f+i_coord_offset,fshift+i_shift_offset);
1456
1457         ggid                        = gid[iidx];
1458         /* Update potential energies */
1459         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
1460         gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
1461
1462         /* Increment number of inner iterations */
1463         inneriter                  += j_index_end - j_index_start;
1464
1465         /* Outer loop uses 20 flops */
1466     }
1467
1468     /* Increment number of outer iterations */
1469     outeriter        += nri;
1470
1471     /* Update outer/inner flops */
1472
1473     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*459);
1474 }
1475 /*
1476  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_sse4_1_single
1477  * Electrostatics interaction: Ewald
1478  * VdW interaction:            LJEwald
1479  * Geometry:                   Water3-Water3
1480  * Calculate force/pot:        Force
1481  */
1482 void
1483 nb_kernel_ElecEwSh_VdwLJEwSh_GeomW3W3_F_sse4_1_single
1484                     (t_nblist                    * gmx_restrict       nlist,
1485                      rvec                        * gmx_restrict          xx,
1486                      rvec                        * gmx_restrict          ff,
1487                      t_forcerec                  * gmx_restrict          fr,
1488                      t_mdatoms                   * gmx_restrict     mdatoms,
1489                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
1490                      t_nrnb                      * gmx_restrict        nrnb)
1491 {
1492     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
1493      * just 0 for non-waters.
1494      * Suffixes A,B,C,D refer to j loop unrolling done with SSE, e.g. for the four different
1495      * jnr indices corresponding to data put in the four positions in the SIMD register.
1496      */
1497     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
1498     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
1499     int              jnrA,jnrB,jnrC,jnrD;
1500     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
1501     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
1502     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
1503     real             rcutoff_scalar;
1504     real             *shiftvec,*fshift,*x,*f;
1505     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
1506     real             scratch[4*DIM];
1507     __m128           tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
1508     int              vdwioffset0;
1509     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
1510     int              vdwioffset1;
1511     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
1512     int              vdwioffset2;
1513     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
1514     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
1515     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
1516     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
1517     __m128           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
1518     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
1519     __m128           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
1520     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
1521     __m128           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
1522     __m128           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
1523     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
1524     __m128           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
1525     __m128           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
1526     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
1527     __m128           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1528     __m128           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1529     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
1530     real             *charge;
1531     int              nvdwtype;
1532     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1533     int              *vdwtype;
1534     real             *vdwparam;
1535     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
1536     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
1537     __m128           c6grid_00;
1538     __m128           c6grid_01;
1539     __m128           c6grid_02;
1540     __m128           c6grid_10;
1541     __m128           c6grid_11;
1542     __m128           c6grid_12;
1543     __m128           c6grid_20;
1544     __m128           c6grid_21;
1545     __m128           c6grid_22;
1546     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
1547     real             *vdwgridparam;
1548     __m128           one_half  = _mm_set1_ps(0.5);
1549     __m128           minus_one = _mm_set1_ps(-1.0);
1550     __m128i          ewitab;
1551     __m128           ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
1552     real             *ewtab;
1553     __m128           dummy_mask,cutoff_mask;
1554     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
1555     __m128           one     = _mm_set1_ps(1.0);
1556     __m128           two     = _mm_set1_ps(2.0);
1557     x                = xx[0];
1558     f                = ff[0];
1559
1560     nri              = nlist->nri;
1561     iinr             = nlist->iinr;
1562     jindex           = nlist->jindex;
1563     jjnr             = nlist->jjnr;
1564     shiftidx         = nlist->shift;
1565     gid              = nlist->gid;
1566     shiftvec         = fr->shift_vec[0];
1567     fshift           = fr->fshift[0];
1568     facel            = _mm_set1_ps(fr->epsfac);
1569     charge           = mdatoms->chargeA;
1570     nvdwtype         = fr->ntype;
1571     vdwparam         = fr->nbfp;
1572     vdwtype          = mdatoms->typeA;
1573     vdwgridparam     = fr->ljpme_c6grid;
1574     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
1575     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
1576     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
1577
1578     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
1579     ewtab            = fr->ic->tabq_coul_F;
1580     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
1581     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
1582
1583     /* Setup water-specific parameters */
1584     inr              = nlist->iinr[0];
1585     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
1586     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
1587     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
1588     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
1589
1590     jq0              = _mm_set1_ps(charge[inr+0]);
1591     jq1              = _mm_set1_ps(charge[inr+1]);
1592     jq2              = _mm_set1_ps(charge[inr+2]);
1593     vdwjidx0A        = 2*vdwtype[inr+0];
1594     qq00             = _mm_mul_ps(iq0,jq0);
1595     c6_00            = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
1596     c12_00           = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
1597     c6grid_00        = _mm_set1_ps(vdwgridparam[vdwioffset0+vdwjidx0A]);
1598     qq01             = _mm_mul_ps(iq0,jq1);
1599     qq02             = _mm_mul_ps(iq0,jq2);
1600     qq10             = _mm_mul_ps(iq1,jq0);
1601     qq11             = _mm_mul_ps(iq1,jq1);
1602     qq12             = _mm_mul_ps(iq1,jq2);
1603     qq20             = _mm_mul_ps(iq2,jq0);
1604     qq21             = _mm_mul_ps(iq2,jq1);
1605     qq22             = _mm_mul_ps(iq2,jq2);
1606
1607     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
1608     rcutoff_scalar   = fr->rcoulomb;
1609     rcutoff          = _mm_set1_ps(rcutoff_scalar);
1610     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
1611
1612     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
1613     rvdw             = _mm_set1_ps(fr->rvdw);
1614
1615     /* Avoid stupid compiler warnings */
1616     jnrA = jnrB = jnrC = jnrD = 0;
1617     j_coord_offsetA = 0;
1618     j_coord_offsetB = 0;
1619     j_coord_offsetC = 0;
1620     j_coord_offsetD = 0;
1621
1622     outeriter        = 0;
1623     inneriter        = 0;
1624
1625     for(iidx=0;iidx<4*DIM;iidx++)
1626     {
1627         scratch[iidx] = 0.0;
1628     }
1629
1630     /* Start outer loop over neighborlists */
1631     for(iidx=0; iidx<nri; iidx++)
1632     {
1633         /* Load shift vector for this list */
1634         i_shift_offset   = DIM*shiftidx[iidx];
1635
1636         /* Load limits for loop over neighbors */
1637         j_index_start    = jindex[iidx];
1638         j_index_end      = jindex[iidx+1];
1639
1640         /* Get outer coordinate index */
1641         inr              = iinr[iidx];
1642         i_coord_offset   = DIM*inr;
1643
1644         /* Load i particle coords and add shift vector */
1645         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1646                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1647
1648         fix0             = _mm_setzero_ps();
1649         fiy0             = _mm_setzero_ps();
1650         fiz0             = _mm_setzero_ps();
1651         fix1             = _mm_setzero_ps();
1652         fiy1             = _mm_setzero_ps();
1653         fiz1             = _mm_setzero_ps();
1654         fix2             = _mm_setzero_ps();
1655         fiy2             = _mm_setzero_ps();
1656         fiz2             = _mm_setzero_ps();
1657
1658         /* Start inner kernel loop */
1659         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1660         {
1661
1662             /* Get j neighbor index, and coordinate index */
1663             jnrA             = jjnr[jidx];
1664             jnrB             = jjnr[jidx+1];
1665             jnrC             = jjnr[jidx+2];
1666             jnrD             = jjnr[jidx+3];
1667             j_coord_offsetA  = DIM*jnrA;
1668             j_coord_offsetB  = DIM*jnrB;
1669             j_coord_offsetC  = DIM*jnrC;
1670             j_coord_offsetD  = DIM*jnrD;
1671
1672             /* load j atom coordinates */
1673             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1674                                               x+j_coord_offsetC,x+j_coord_offsetD,
1675                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1676
1677             /* Calculate displacement vector */
1678             dx00             = _mm_sub_ps(ix0,jx0);
1679             dy00             = _mm_sub_ps(iy0,jy0);
1680             dz00             = _mm_sub_ps(iz0,jz0);
1681             dx01             = _mm_sub_ps(ix0,jx1);
1682             dy01             = _mm_sub_ps(iy0,jy1);
1683             dz01             = _mm_sub_ps(iz0,jz1);
1684             dx02             = _mm_sub_ps(ix0,jx2);
1685             dy02             = _mm_sub_ps(iy0,jy2);
1686             dz02             = _mm_sub_ps(iz0,jz2);
1687             dx10             = _mm_sub_ps(ix1,jx0);
1688             dy10             = _mm_sub_ps(iy1,jy0);
1689             dz10             = _mm_sub_ps(iz1,jz0);
1690             dx11             = _mm_sub_ps(ix1,jx1);
1691             dy11             = _mm_sub_ps(iy1,jy1);
1692             dz11             = _mm_sub_ps(iz1,jz1);
1693             dx12             = _mm_sub_ps(ix1,jx2);
1694             dy12             = _mm_sub_ps(iy1,jy2);
1695             dz12             = _mm_sub_ps(iz1,jz2);
1696             dx20             = _mm_sub_ps(ix2,jx0);
1697             dy20             = _mm_sub_ps(iy2,jy0);
1698             dz20             = _mm_sub_ps(iz2,jz0);
1699             dx21             = _mm_sub_ps(ix2,jx1);
1700             dy21             = _mm_sub_ps(iy2,jy1);
1701             dz21             = _mm_sub_ps(iz2,jz1);
1702             dx22             = _mm_sub_ps(ix2,jx2);
1703             dy22             = _mm_sub_ps(iy2,jy2);
1704             dz22             = _mm_sub_ps(iz2,jz2);
1705
1706             /* Calculate squared distance and things based on it */
1707             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1708             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1709             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1710             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1711             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1712             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1713             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1714             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1715             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1716
1717             rinv00           = gmx_mm_invsqrt_ps(rsq00);
1718             rinv01           = gmx_mm_invsqrt_ps(rsq01);
1719             rinv02           = gmx_mm_invsqrt_ps(rsq02);
1720             rinv10           = gmx_mm_invsqrt_ps(rsq10);
1721             rinv11           = gmx_mm_invsqrt_ps(rsq11);
1722             rinv12           = gmx_mm_invsqrt_ps(rsq12);
1723             rinv20           = gmx_mm_invsqrt_ps(rsq20);
1724             rinv21           = gmx_mm_invsqrt_ps(rsq21);
1725             rinv22           = gmx_mm_invsqrt_ps(rsq22);
1726
1727             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1728             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
1729             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
1730             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1731             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
1732             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
1733             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1734             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
1735             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
1736
1737             fjx0             = _mm_setzero_ps();
1738             fjy0             = _mm_setzero_ps();
1739             fjz0             = _mm_setzero_ps();
1740             fjx1             = _mm_setzero_ps();
1741             fjy1             = _mm_setzero_ps();
1742             fjz1             = _mm_setzero_ps();
1743             fjx2             = _mm_setzero_ps();
1744             fjy2             = _mm_setzero_ps();
1745             fjz2             = _mm_setzero_ps();
1746
1747             /**************************
1748              * CALCULATE INTERACTIONS *
1749              **************************/
1750
1751             if (gmx_mm_any_lt(rsq00,rcutoff2))
1752             {
1753
1754             r00              = _mm_mul_ps(rsq00,rinv00);
1755
1756             /* EWALD ELECTROSTATICS */
1757
1758             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1759             ewrt             = _mm_mul_ps(r00,ewtabscale);
1760             ewitab           = _mm_cvttps_epi32(ewrt);
1761             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1762             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1763                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1764                                          &ewtabF,&ewtabFn);
1765             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1766             felec            = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
1767
1768             /* Analytical LJ-PME */
1769             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
1770             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
1771             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
1772             exponent         = gmx_simd_exp_r(ewcljrsq);
1773             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
1774             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
1775             /* f6A = 6 * C6grid * (1 - poly) */
1776             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
1777             /* f6B = C6grid * exponent * beta^6 */
1778             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
1779             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
1780             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);
1781
1782             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
1783
1784             fscal            = _mm_add_ps(felec,fvdw);
1785
1786             fscal            = _mm_and_ps(fscal,cutoff_mask);
1787
1788             /* Calculate temporary vectorial force */
1789             tx               = _mm_mul_ps(fscal,dx00);
1790             ty               = _mm_mul_ps(fscal,dy00);
1791             tz               = _mm_mul_ps(fscal,dz00);
1792
1793             /* Update vectorial force */
1794             fix0             = _mm_add_ps(fix0,tx);
1795             fiy0             = _mm_add_ps(fiy0,ty);
1796             fiz0             = _mm_add_ps(fiz0,tz);
1797
1798             fjx0             = _mm_add_ps(fjx0,tx);
1799             fjy0             = _mm_add_ps(fjy0,ty);
1800             fjz0             = _mm_add_ps(fjz0,tz);
1801
1802             }
1803
1804             /**************************
1805              * CALCULATE INTERACTIONS *
1806              **************************/
1807
1808             if (gmx_mm_any_lt(rsq01,rcutoff2))
1809             {
1810
1811             r01              = _mm_mul_ps(rsq01,rinv01);
1812
1813             /* EWALD ELECTROSTATICS */
1814
1815             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1816             ewrt             = _mm_mul_ps(r01,ewtabscale);
1817             ewitab           = _mm_cvttps_epi32(ewrt);
1818             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1819             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1820                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1821                                          &ewtabF,&ewtabFn);
1822             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1823             felec            = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec));
1824
1825             cutoff_mask      = _mm_cmplt_ps(rsq01,rcutoff2);
1826
1827             fscal            = felec;
1828
1829             fscal            = _mm_and_ps(fscal,cutoff_mask);
1830
1831             /* Calculate temporary vectorial force */
1832             tx               = _mm_mul_ps(fscal,dx01);
1833             ty               = _mm_mul_ps(fscal,dy01);
1834             tz               = _mm_mul_ps(fscal,dz01);
1835
1836             /* Update vectorial force */
1837             fix0             = _mm_add_ps(fix0,tx);
1838             fiy0             = _mm_add_ps(fiy0,ty);
1839             fiz0             = _mm_add_ps(fiz0,tz);
1840
1841             fjx1             = _mm_add_ps(fjx1,tx);
1842             fjy1             = _mm_add_ps(fjy1,ty);
1843             fjz1             = _mm_add_ps(fjz1,tz);
1844
1845             }
1846
1847             /**************************
1848              * CALCULATE INTERACTIONS *
1849              **************************/
1850
1851             if (gmx_mm_any_lt(rsq02,rcutoff2))
1852             {
1853
1854             r02              = _mm_mul_ps(rsq02,rinv02);
1855
1856             /* EWALD ELECTROSTATICS */
1857
1858             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1859             ewrt             = _mm_mul_ps(r02,ewtabscale);
1860             ewitab           = _mm_cvttps_epi32(ewrt);
1861             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1862             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1863                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1864                                          &ewtabF,&ewtabFn);
1865             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1866             felec            = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec));
1867
1868             cutoff_mask      = _mm_cmplt_ps(rsq02,rcutoff2);
1869
1870             fscal            = felec;
1871
1872             fscal            = _mm_and_ps(fscal,cutoff_mask);
1873
1874             /* Calculate temporary vectorial force */
1875             tx               = _mm_mul_ps(fscal,dx02);
1876             ty               = _mm_mul_ps(fscal,dy02);
1877             tz               = _mm_mul_ps(fscal,dz02);
1878
1879             /* Update vectorial force */
1880             fix0             = _mm_add_ps(fix0,tx);
1881             fiy0             = _mm_add_ps(fiy0,ty);
1882             fiz0             = _mm_add_ps(fiz0,tz);
1883
1884             fjx2             = _mm_add_ps(fjx2,tx);
1885             fjy2             = _mm_add_ps(fjy2,ty);
1886             fjz2             = _mm_add_ps(fjz2,tz);
1887
1888             }
1889
1890             /**************************
1891              * CALCULATE INTERACTIONS *
1892              **************************/
1893
1894             if (gmx_mm_any_lt(rsq10,rcutoff2))
1895             {
1896
1897             r10              = _mm_mul_ps(rsq10,rinv10);
1898
1899             /* EWALD ELECTROSTATICS */
1900
1901             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1902             ewrt             = _mm_mul_ps(r10,ewtabscale);
1903             ewitab           = _mm_cvttps_epi32(ewrt);
1904             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1905             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1906                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1907                                          &ewtabF,&ewtabFn);
1908             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1909             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
1910
1911             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1912
1913             fscal            = felec;
1914
1915             fscal            = _mm_and_ps(fscal,cutoff_mask);
1916
1917             /* Calculate temporary vectorial force */
1918             tx               = _mm_mul_ps(fscal,dx10);
1919             ty               = _mm_mul_ps(fscal,dy10);
1920             tz               = _mm_mul_ps(fscal,dz10);
1921
1922             /* Update vectorial force */
1923             fix1             = _mm_add_ps(fix1,tx);
1924             fiy1             = _mm_add_ps(fiy1,ty);
1925             fiz1             = _mm_add_ps(fiz1,tz);
1926
1927             fjx0             = _mm_add_ps(fjx0,tx);
1928             fjy0             = _mm_add_ps(fjy0,ty);
1929             fjz0             = _mm_add_ps(fjz0,tz);
1930
1931             }
1932
1933             /**************************
1934              * CALCULATE INTERACTIONS *
1935              **************************/
1936
1937             if (gmx_mm_any_lt(rsq11,rcutoff2))
1938             {
1939
1940             r11              = _mm_mul_ps(rsq11,rinv11);
1941
1942             /* EWALD ELECTROSTATICS */
1943
1944             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1945             ewrt             = _mm_mul_ps(r11,ewtabscale);
1946             ewitab           = _mm_cvttps_epi32(ewrt);
1947             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1948             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1949                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1950                                          &ewtabF,&ewtabFn);
1951             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1952             felec            = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
1953
1954             cutoff_mask      = _mm_cmplt_ps(rsq11,rcutoff2);
1955
1956             fscal            = felec;
1957
1958             fscal            = _mm_and_ps(fscal,cutoff_mask);
1959
1960             /* Calculate temporary vectorial force */
1961             tx               = _mm_mul_ps(fscal,dx11);
1962             ty               = _mm_mul_ps(fscal,dy11);
1963             tz               = _mm_mul_ps(fscal,dz11);
1964
1965             /* Update vectorial force */
1966             fix1             = _mm_add_ps(fix1,tx);
1967             fiy1             = _mm_add_ps(fiy1,ty);
1968             fiz1             = _mm_add_ps(fiz1,tz);
1969
1970             fjx1             = _mm_add_ps(fjx1,tx);
1971             fjy1             = _mm_add_ps(fjy1,ty);
1972             fjz1             = _mm_add_ps(fjz1,tz);
1973
1974             }
1975
1976             /**************************
1977              * CALCULATE INTERACTIONS *
1978              **************************/
1979
1980             if (gmx_mm_any_lt(rsq12,rcutoff2))
1981             {
1982
1983             r12              = _mm_mul_ps(rsq12,rinv12);
1984
1985             /* EWALD ELECTROSTATICS */
1986
1987             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1988             ewrt             = _mm_mul_ps(r12,ewtabscale);
1989             ewitab           = _mm_cvttps_epi32(ewrt);
1990             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
1991             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
1992                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
1993                                          &ewtabF,&ewtabFn);
1994             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
1995             felec            = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
1996
1997             cutoff_mask      = _mm_cmplt_ps(rsq12,rcutoff2);
1998
1999             fscal            = felec;
2000
2001             fscal            = _mm_and_ps(fscal,cutoff_mask);
2002
2003             /* Calculate temporary vectorial force */
2004             tx               = _mm_mul_ps(fscal,dx12);
2005             ty               = _mm_mul_ps(fscal,dy12);
2006             tz               = _mm_mul_ps(fscal,dz12);
2007
2008             /* Update vectorial force */
2009             fix1             = _mm_add_ps(fix1,tx);
2010             fiy1             = _mm_add_ps(fiy1,ty);
2011             fiz1             = _mm_add_ps(fiz1,tz);
2012
2013             fjx2             = _mm_add_ps(fjx2,tx);
2014             fjy2             = _mm_add_ps(fjy2,ty);
2015             fjz2             = _mm_add_ps(fjz2,tz);
2016
2017             }
2018
2019             /**************************
2020              * CALCULATE INTERACTIONS *
2021              **************************/
2022
2023             if (gmx_mm_any_lt(rsq20,rcutoff2))
2024             {
2025
2026             r20              = _mm_mul_ps(rsq20,rinv20);
2027
2028             /* EWALD ELECTROSTATICS */
2029
2030             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2031             ewrt             = _mm_mul_ps(r20,ewtabscale);
2032             ewitab           = _mm_cvttps_epi32(ewrt);
2033             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2034             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2035                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2036                                          &ewtabF,&ewtabFn);
2037             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2038             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
2039
2040             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
2041
2042             fscal            = felec;
2043
2044             fscal            = _mm_and_ps(fscal,cutoff_mask);
2045
2046             /* Calculate temporary vectorial force */
2047             tx               = _mm_mul_ps(fscal,dx20);
2048             ty               = _mm_mul_ps(fscal,dy20);
2049             tz               = _mm_mul_ps(fscal,dz20);
2050
2051             /* Update vectorial force */
2052             fix2             = _mm_add_ps(fix2,tx);
2053             fiy2             = _mm_add_ps(fiy2,ty);
2054             fiz2             = _mm_add_ps(fiz2,tz);
2055
2056             fjx0             = _mm_add_ps(fjx0,tx);
2057             fjy0             = _mm_add_ps(fjy0,ty);
2058             fjz0             = _mm_add_ps(fjz0,tz);
2059
2060             }
2061
2062             /**************************
2063              * CALCULATE INTERACTIONS *
2064              **************************/
2065
2066             if (gmx_mm_any_lt(rsq21,rcutoff2))
2067             {
2068
2069             r21              = _mm_mul_ps(rsq21,rinv21);
2070
2071             /* EWALD ELECTROSTATICS */
2072
2073             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2074             ewrt             = _mm_mul_ps(r21,ewtabscale);
2075             ewitab           = _mm_cvttps_epi32(ewrt);
2076             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2077             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2078                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2079                                          &ewtabF,&ewtabFn);
2080             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2081             felec            = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
2082
2083             cutoff_mask      = _mm_cmplt_ps(rsq21,rcutoff2);
2084
2085             fscal            = felec;
2086
2087             fscal            = _mm_and_ps(fscal,cutoff_mask);
2088
2089             /* Calculate temporary vectorial force */
2090             tx               = _mm_mul_ps(fscal,dx21);
2091             ty               = _mm_mul_ps(fscal,dy21);
2092             tz               = _mm_mul_ps(fscal,dz21);
2093
2094             /* Update vectorial force */
2095             fix2             = _mm_add_ps(fix2,tx);
2096             fiy2             = _mm_add_ps(fiy2,ty);
2097             fiz2             = _mm_add_ps(fiz2,tz);
2098
2099             fjx1             = _mm_add_ps(fjx1,tx);
2100             fjy1             = _mm_add_ps(fjy1,ty);
2101             fjz1             = _mm_add_ps(fjz1,tz);
2102
2103             }
2104
2105             /**************************
2106              * CALCULATE INTERACTIONS *
2107              **************************/
2108
2109             if (gmx_mm_any_lt(rsq22,rcutoff2))
2110             {
2111
2112             r22              = _mm_mul_ps(rsq22,rinv22);
2113
2114             /* EWALD ELECTROSTATICS */
2115
2116             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2117             ewrt             = _mm_mul_ps(r22,ewtabscale);
2118             ewitab           = _mm_cvttps_epi32(ewrt);
2119             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2120             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2121                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2122                                          &ewtabF,&ewtabFn);
2123             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2124             felec            = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
2125
2126             cutoff_mask      = _mm_cmplt_ps(rsq22,rcutoff2);
2127
2128             fscal            = felec;
2129
2130             fscal            = _mm_and_ps(fscal,cutoff_mask);
2131
2132             /* Calculate temporary vectorial force */
2133             tx               = _mm_mul_ps(fscal,dx22);
2134             ty               = _mm_mul_ps(fscal,dy22);
2135             tz               = _mm_mul_ps(fscal,dz22);
2136
2137             /* Update vectorial force */
2138             fix2             = _mm_add_ps(fix2,tx);
2139             fiy2             = _mm_add_ps(fiy2,ty);
2140             fiz2             = _mm_add_ps(fiz2,tz);
2141
2142             fjx2             = _mm_add_ps(fjx2,tx);
2143             fjy2             = _mm_add_ps(fjy2,ty);
2144             fjz2             = _mm_add_ps(fjz2,tz);
2145
2146             }
2147
2148             fjptrA             = f+j_coord_offsetA;
2149             fjptrB             = f+j_coord_offsetB;
2150             fjptrC             = f+j_coord_offsetC;
2151             fjptrD             = f+j_coord_offsetD;
2152
2153             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
2154                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2155
2156             /* Inner loop uses 374 flops */
2157         }
2158
2159         if(jidx<j_index_end)
2160         {
2161
2162             /* Get j neighbor index, and coordinate index */
2163             jnrlistA         = jjnr[jidx];
2164             jnrlistB         = jjnr[jidx+1];
2165             jnrlistC         = jjnr[jidx+2];
2166             jnrlistD         = jjnr[jidx+3];
2167             /* Sign of each element will be negative for non-real atoms.
2168              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
2169              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
2170              */
2171             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
2172             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
2173             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
2174             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
2175             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
2176             j_coord_offsetA  = DIM*jnrA;
2177             j_coord_offsetB  = DIM*jnrB;
2178             j_coord_offsetC  = DIM*jnrC;
2179             j_coord_offsetD  = DIM*jnrD;
2180
2181             /* load j atom coordinates */
2182             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
2183                                               x+j_coord_offsetC,x+j_coord_offsetD,
2184                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
2185
2186             /* Calculate displacement vector */
2187             dx00             = _mm_sub_ps(ix0,jx0);
2188             dy00             = _mm_sub_ps(iy0,jy0);
2189             dz00             = _mm_sub_ps(iz0,jz0);
2190             dx01             = _mm_sub_ps(ix0,jx1);
2191             dy01             = _mm_sub_ps(iy0,jy1);
2192             dz01             = _mm_sub_ps(iz0,jz1);
2193             dx02             = _mm_sub_ps(ix0,jx2);
2194             dy02             = _mm_sub_ps(iy0,jy2);
2195             dz02             = _mm_sub_ps(iz0,jz2);
2196             dx10             = _mm_sub_ps(ix1,jx0);
2197             dy10             = _mm_sub_ps(iy1,jy0);
2198             dz10             = _mm_sub_ps(iz1,jz0);
2199             dx11             = _mm_sub_ps(ix1,jx1);
2200             dy11             = _mm_sub_ps(iy1,jy1);
2201             dz11             = _mm_sub_ps(iz1,jz1);
2202             dx12             = _mm_sub_ps(ix1,jx2);
2203             dy12             = _mm_sub_ps(iy1,jy2);
2204             dz12             = _mm_sub_ps(iz1,jz2);
2205             dx20             = _mm_sub_ps(ix2,jx0);
2206             dy20             = _mm_sub_ps(iy2,jy0);
2207             dz20             = _mm_sub_ps(iz2,jz0);
2208             dx21             = _mm_sub_ps(ix2,jx1);
2209             dy21             = _mm_sub_ps(iy2,jy1);
2210             dz21             = _mm_sub_ps(iz2,jz1);
2211             dx22             = _mm_sub_ps(ix2,jx2);
2212             dy22             = _mm_sub_ps(iy2,jy2);
2213             dz22             = _mm_sub_ps(iz2,jz2);
2214
2215             /* Calculate squared distance and things based on it */
2216             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
2217             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
2218             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
2219             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
2220             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
2221             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
2222             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
2223             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
2224             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
2225
2226             rinv00           = gmx_mm_invsqrt_ps(rsq00);
2227             rinv01           = gmx_mm_invsqrt_ps(rsq01);
2228             rinv02           = gmx_mm_invsqrt_ps(rsq02);
2229             rinv10           = gmx_mm_invsqrt_ps(rsq10);
2230             rinv11           = gmx_mm_invsqrt_ps(rsq11);
2231             rinv12           = gmx_mm_invsqrt_ps(rsq12);
2232             rinv20           = gmx_mm_invsqrt_ps(rsq20);
2233             rinv21           = gmx_mm_invsqrt_ps(rsq21);
2234             rinv22           = gmx_mm_invsqrt_ps(rsq22);
2235
2236             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
2237             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
2238             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
2239             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
2240             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
2241             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
2242             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
2243             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
2244             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
2245
2246             fjx0             = _mm_setzero_ps();
2247             fjy0             = _mm_setzero_ps();
2248             fjz0             = _mm_setzero_ps();
2249             fjx1             = _mm_setzero_ps();
2250             fjy1             = _mm_setzero_ps();
2251             fjz1             = _mm_setzero_ps();
2252             fjx2             = _mm_setzero_ps();
2253             fjy2             = _mm_setzero_ps();
2254             fjz2             = _mm_setzero_ps();
2255
2256             /**************************
2257              * CALCULATE INTERACTIONS *
2258              **************************/
2259
2260             if (gmx_mm_any_lt(rsq00,rcutoff2))
2261             {
2262
2263             r00              = _mm_mul_ps(rsq00,rinv00);
2264             r00              = _mm_andnot_ps(dummy_mask,r00);
2265
2266             /* EWALD ELECTROSTATICS */
2267
2268             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2269             ewrt             = _mm_mul_ps(r00,ewtabscale);
2270             ewitab           = _mm_cvttps_epi32(ewrt);
2271             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2272             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2273                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2274                                          &ewtabF,&ewtabFn);
2275             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2276             felec            = _mm_mul_ps(_mm_mul_ps(qq00,rinv00),_mm_sub_ps(rinvsq00,felec));
2277
2278             /* Analytical LJ-PME */
2279             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
2280             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
2281             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
2282             exponent         = gmx_simd_exp_r(ewcljrsq);
2283             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
2284             poly             = _mm_mul_ps(exponent,_mm_add_ps(_mm_sub_ps(one,ewcljrsq),_mm_mul_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half)));
2285             /* f6A = 6 * C6grid * (1 - poly) */
2286             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
2287             /* f6B = C6grid * exponent * beta^6 */
2288             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
2289             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
2290             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);
2291
2292             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
2293
2294             fscal            = _mm_add_ps(felec,fvdw);
2295
2296             fscal            = _mm_and_ps(fscal,cutoff_mask);
2297
2298             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2299
2300             /* Calculate temporary vectorial force */
2301             tx               = _mm_mul_ps(fscal,dx00);
2302             ty               = _mm_mul_ps(fscal,dy00);
2303             tz               = _mm_mul_ps(fscal,dz00);
2304
2305             /* Update vectorial force */
2306             fix0             = _mm_add_ps(fix0,tx);
2307             fiy0             = _mm_add_ps(fiy0,ty);
2308             fiz0             = _mm_add_ps(fiz0,tz);
2309
2310             fjx0             = _mm_add_ps(fjx0,tx);
2311             fjy0             = _mm_add_ps(fjy0,ty);
2312             fjz0             = _mm_add_ps(fjz0,tz);
2313
2314             }
2315
2316             /**************************
2317              * CALCULATE INTERACTIONS *
2318              **************************/
2319
2320             if (gmx_mm_any_lt(rsq01,rcutoff2))
2321             {
2322
2323             r01              = _mm_mul_ps(rsq01,rinv01);
2324             r01              = _mm_andnot_ps(dummy_mask,r01);
2325
2326             /* EWALD ELECTROSTATICS */
2327
2328             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2329             ewrt             = _mm_mul_ps(r01,ewtabscale);
2330             ewitab           = _mm_cvttps_epi32(ewrt);
2331             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2332             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2333                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2334                                          &ewtabF,&ewtabFn);
2335             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2336             felec            = _mm_mul_ps(_mm_mul_ps(qq01,rinv01),_mm_sub_ps(rinvsq01,felec));
2337
2338             cutoff_mask      = _mm_cmplt_ps(rsq01,rcutoff2);
2339
2340             fscal            = felec;
2341
2342             fscal            = _mm_and_ps(fscal,cutoff_mask);
2343
2344             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2345
2346             /* Calculate temporary vectorial force */
2347             tx               = _mm_mul_ps(fscal,dx01);
2348             ty               = _mm_mul_ps(fscal,dy01);
2349             tz               = _mm_mul_ps(fscal,dz01);
2350
2351             /* Update vectorial force */
2352             fix0             = _mm_add_ps(fix0,tx);
2353             fiy0             = _mm_add_ps(fiy0,ty);
2354             fiz0             = _mm_add_ps(fiz0,tz);
2355
2356             fjx1             = _mm_add_ps(fjx1,tx);
2357             fjy1             = _mm_add_ps(fjy1,ty);
2358             fjz1             = _mm_add_ps(fjz1,tz);
2359
2360             }
2361
2362             /**************************
2363              * CALCULATE INTERACTIONS *
2364              **************************/
2365
2366             if (gmx_mm_any_lt(rsq02,rcutoff2))
2367             {
2368
2369             r02              = _mm_mul_ps(rsq02,rinv02);
2370             r02              = _mm_andnot_ps(dummy_mask,r02);
2371
2372             /* EWALD ELECTROSTATICS */
2373
2374             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2375             ewrt             = _mm_mul_ps(r02,ewtabscale);
2376             ewitab           = _mm_cvttps_epi32(ewrt);
2377             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2378             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2379                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2380                                          &ewtabF,&ewtabFn);
2381             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2382             felec            = _mm_mul_ps(_mm_mul_ps(qq02,rinv02),_mm_sub_ps(rinvsq02,felec));
2383
2384             cutoff_mask      = _mm_cmplt_ps(rsq02,rcutoff2);
2385
2386             fscal            = felec;
2387
2388             fscal            = _mm_and_ps(fscal,cutoff_mask);
2389
2390             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2391
2392             /* Calculate temporary vectorial force */
2393             tx               = _mm_mul_ps(fscal,dx02);
2394             ty               = _mm_mul_ps(fscal,dy02);
2395             tz               = _mm_mul_ps(fscal,dz02);
2396
2397             /* Update vectorial force */
2398             fix0             = _mm_add_ps(fix0,tx);
2399             fiy0             = _mm_add_ps(fiy0,ty);
2400             fiz0             = _mm_add_ps(fiz0,tz);
2401
2402             fjx2             = _mm_add_ps(fjx2,tx);
2403             fjy2             = _mm_add_ps(fjy2,ty);
2404             fjz2             = _mm_add_ps(fjz2,tz);
2405
2406             }
2407
2408             /**************************
2409              * CALCULATE INTERACTIONS *
2410              **************************/
2411
2412             if (gmx_mm_any_lt(rsq10,rcutoff2))
2413             {
2414
2415             r10              = _mm_mul_ps(rsq10,rinv10);
2416             r10              = _mm_andnot_ps(dummy_mask,r10);
2417
2418             /* EWALD ELECTROSTATICS */
2419
2420             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2421             ewrt             = _mm_mul_ps(r10,ewtabscale);
2422             ewitab           = _mm_cvttps_epi32(ewrt);
2423             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2424             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2425                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2426                                          &ewtabF,&ewtabFn);
2427             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2428             felec            = _mm_mul_ps(_mm_mul_ps(qq10,rinv10),_mm_sub_ps(rinvsq10,felec));
2429
2430             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
2431
2432             fscal            = felec;
2433
2434             fscal            = _mm_and_ps(fscal,cutoff_mask);
2435
2436             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2437
2438             /* Calculate temporary vectorial force */
2439             tx               = _mm_mul_ps(fscal,dx10);
2440             ty               = _mm_mul_ps(fscal,dy10);
2441             tz               = _mm_mul_ps(fscal,dz10);
2442
2443             /* Update vectorial force */
2444             fix1             = _mm_add_ps(fix1,tx);
2445             fiy1             = _mm_add_ps(fiy1,ty);
2446             fiz1             = _mm_add_ps(fiz1,tz);
2447
2448             fjx0             = _mm_add_ps(fjx0,tx);
2449             fjy0             = _mm_add_ps(fjy0,ty);
2450             fjz0             = _mm_add_ps(fjz0,tz);
2451
2452             }
2453
2454             /**************************
2455              * CALCULATE INTERACTIONS *
2456              **************************/
2457
2458             if (gmx_mm_any_lt(rsq11,rcutoff2))
2459             {
2460
2461             r11              = _mm_mul_ps(rsq11,rinv11);
2462             r11              = _mm_andnot_ps(dummy_mask,r11);
2463
2464             /* EWALD ELECTROSTATICS */
2465
2466             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2467             ewrt             = _mm_mul_ps(r11,ewtabscale);
2468             ewitab           = _mm_cvttps_epi32(ewrt);
2469             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2470             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2471                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2472                                          &ewtabF,&ewtabFn);
2473             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2474             felec            = _mm_mul_ps(_mm_mul_ps(qq11,rinv11),_mm_sub_ps(rinvsq11,felec));
2475
2476             cutoff_mask      = _mm_cmplt_ps(rsq11,rcutoff2);
2477
2478             fscal            = felec;
2479
2480             fscal            = _mm_and_ps(fscal,cutoff_mask);
2481
2482             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2483
2484             /* Calculate temporary vectorial force */
2485             tx               = _mm_mul_ps(fscal,dx11);
2486             ty               = _mm_mul_ps(fscal,dy11);
2487             tz               = _mm_mul_ps(fscal,dz11);
2488
2489             /* Update vectorial force */
2490             fix1             = _mm_add_ps(fix1,tx);
2491             fiy1             = _mm_add_ps(fiy1,ty);
2492             fiz1             = _mm_add_ps(fiz1,tz);
2493
2494             fjx1             = _mm_add_ps(fjx1,tx);
2495             fjy1             = _mm_add_ps(fjy1,ty);
2496             fjz1             = _mm_add_ps(fjz1,tz);
2497
2498             }
2499
2500             /**************************
2501              * CALCULATE INTERACTIONS *
2502              **************************/
2503
2504             if (gmx_mm_any_lt(rsq12,rcutoff2))
2505             {
2506
2507             r12              = _mm_mul_ps(rsq12,rinv12);
2508             r12              = _mm_andnot_ps(dummy_mask,r12);
2509
2510             /* EWALD ELECTROSTATICS */
2511
2512             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2513             ewrt             = _mm_mul_ps(r12,ewtabscale);
2514             ewitab           = _mm_cvttps_epi32(ewrt);
2515             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2516             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2517                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2518                                          &ewtabF,&ewtabFn);
2519             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2520             felec            = _mm_mul_ps(_mm_mul_ps(qq12,rinv12),_mm_sub_ps(rinvsq12,felec));
2521
2522             cutoff_mask      = _mm_cmplt_ps(rsq12,rcutoff2);
2523
2524             fscal            = felec;
2525
2526             fscal            = _mm_and_ps(fscal,cutoff_mask);
2527
2528             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2529
2530             /* Calculate temporary vectorial force */
2531             tx               = _mm_mul_ps(fscal,dx12);
2532             ty               = _mm_mul_ps(fscal,dy12);
2533             tz               = _mm_mul_ps(fscal,dz12);
2534
2535             /* Update vectorial force */
2536             fix1             = _mm_add_ps(fix1,tx);
2537             fiy1             = _mm_add_ps(fiy1,ty);
2538             fiz1             = _mm_add_ps(fiz1,tz);
2539
2540             fjx2             = _mm_add_ps(fjx2,tx);
2541             fjy2             = _mm_add_ps(fjy2,ty);
2542             fjz2             = _mm_add_ps(fjz2,tz);
2543
2544             }
2545
2546             /**************************
2547              * CALCULATE INTERACTIONS *
2548              **************************/
2549
2550             if (gmx_mm_any_lt(rsq20,rcutoff2))
2551             {
2552
2553             r20              = _mm_mul_ps(rsq20,rinv20);
2554             r20              = _mm_andnot_ps(dummy_mask,r20);
2555
2556             /* EWALD ELECTROSTATICS */
2557
2558             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2559             ewrt             = _mm_mul_ps(r20,ewtabscale);
2560             ewitab           = _mm_cvttps_epi32(ewrt);
2561             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2562             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2563                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2564                                          &ewtabF,&ewtabFn);
2565             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2566             felec            = _mm_mul_ps(_mm_mul_ps(qq20,rinv20),_mm_sub_ps(rinvsq20,felec));
2567
2568             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
2569
2570             fscal            = felec;
2571
2572             fscal            = _mm_and_ps(fscal,cutoff_mask);
2573
2574             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2575
2576             /* Calculate temporary vectorial force */
2577             tx               = _mm_mul_ps(fscal,dx20);
2578             ty               = _mm_mul_ps(fscal,dy20);
2579             tz               = _mm_mul_ps(fscal,dz20);
2580
2581             /* Update vectorial force */
2582             fix2             = _mm_add_ps(fix2,tx);
2583             fiy2             = _mm_add_ps(fiy2,ty);
2584             fiz2             = _mm_add_ps(fiz2,tz);
2585
2586             fjx0             = _mm_add_ps(fjx0,tx);
2587             fjy0             = _mm_add_ps(fjy0,ty);
2588             fjz0             = _mm_add_ps(fjz0,tz);
2589
2590             }
2591
2592             /**************************
2593              * CALCULATE INTERACTIONS *
2594              **************************/
2595
2596             if (gmx_mm_any_lt(rsq21,rcutoff2))
2597             {
2598
2599             r21              = _mm_mul_ps(rsq21,rinv21);
2600             r21              = _mm_andnot_ps(dummy_mask,r21);
2601
2602             /* EWALD ELECTROSTATICS */
2603
2604             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2605             ewrt             = _mm_mul_ps(r21,ewtabscale);
2606             ewitab           = _mm_cvttps_epi32(ewrt);
2607             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2608             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2609                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2610                                          &ewtabF,&ewtabFn);
2611             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2612             felec            = _mm_mul_ps(_mm_mul_ps(qq21,rinv21),_mm_sub_ps(rinvsq21,felec));
2613
2614             cutoff_mask      = _mm_cmplt_ps(rsq21,rcutoff2);
2615
2616             fscal            = felec;
2617
2618             fscal            = _mm_and_ps(fscal,cutoff_mask);
2619
2620             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2621
2622             /* Calculate temporary vectorial force */
2623             tx               = _mm_mul_ps(fscal,dx21);
2624             ty               = _mm_mul_ps(fscal,dy21);
2625             tz               = _mm_mul_ps(fscal,dz21);
2626
2627             /* Update vectorial force */
2628             fix2             = _mm_add_ps(fix2,tx);
2629             fiy2             = _mm_add_ps(fiy2,ty);
2630             fiz2             = _mm_add_ps(fiz2,tz);
2631
2632             fjx1             = _mm_add_ps(fjx1,tx);
2633             fjy1             = _mm_add_ps(fjy1,ty);
2634             fjz1             = _mm_add_ps(fjz1,tz);
2635
2636             }
2637
2638             /**************************
2639              * CALCULATE INTERACTIONS *
2640              **************************/
2641
2642             if (gmx_mm_any_lt(rsq22,rcutoff2))
2643             {
2644
2645             r22              = _mm_mul_ps(rsq22,rinv22);
2646             r22              = _mm_andnot_ps(dummy_mask,r22);
2647
2648             /* EWALD ELECTROSTATICS */
2649
2650             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
2651             ewrt             = _mm_mul_ps(r22,ewtabscale);
2652             ewitab           = _mm_cvttps_epi32(ewrt);
2653             eweps            = _mm_sub_ps(ewrt,_mm_round_ps(ewrt, _MM_FROUND_FLOOR));
2654             gmx_mm_load_4pair_swizzle_ps(ewtab + gmx_mm_extract_epi32(ewitab,0),ewtab + gmx_mm_extract_epi32(ewitab,1),
2655                                          ewtab + gmx_mm_extract_epi32(ewitab,2),ewtab + gmx_mm_extract_epi32(ewitab,3),
2656                                          &ewtabF,&ewtabFn);
2657             felec            = _mm_add_ps(_mm_mul_ps( _mm_sub_ps(one,eweps),ewtabF),_mm_mul_ps(eweps,ewtabFn));
2658             felec            = _mm_mul_ps(_mm_mul_ps(qq22,rinv22),_mm_sub_ps(rinvsq22,felec));
2659
2660             cutoff_mask      = _mm_cmplt_ps(rsq22,rcutoff2);
2661
2662             fscal            = felec;
2663
2664             fscal            = _mm_and_ps(fscal,cutoff_mask);
2665
2666             fscal            = _mm_andnot_ps(dummy_mask,fscal);
2667
2668             /* Calculate temporary vectorial force */
2669             tx               = _mm_mul_ps(fscal,dx22);
2670             ty               = _mm_mul_ps(fscal,dy22);
2671             tz               = _mm_mul_ps(fscal,dz22);
2672
2673             /* Update vectorial force */
2674             fix2             = _mm_add_ps(fix2,tx);
2675             fiy2             = _mm_add_ps(fiy2,ty);
2676             fiz2             = _mm_add_ps(fiz2,tz);
2677
2678             fjx2             = _mm_add_ps(fjx2,tx);
2679             fjy2             = _mm_add_ps(fjy2,ty);
2680             fjz2             = _mm_add_ps(fjz2,tz);
2681
2682             }
2683
2684             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
2685             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
2686             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
2687             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
2688
2689             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
2690                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
2691
2692             /* Inner loop uses 383 flops */
2693         }
2694
2695         /* End of innermost loop */
2696
2697         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
2698                                               f+i_coord_offset,fshift+i_shift_offset);
2699
2700         /* Increment number of inner iterations */
2701         inneriter                  += j_index_end - j_index_start;
2702
2703         /* Outer loop uses 18 flops */
2704     }
2705
2706     /* Increment number of outer iterations */
2707     outeriter        += nri;
2708
2709     /* Update outer/inner flops */
2710
2711     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*383);
2712 }