src/gromacs/gmxlib/nonbonded/nb_kernel_avx_128_fma_single/nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_avx_128_fma_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2017, 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 avx_128_fma_single kernel generator.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/gmxlib/nrnb.h"
  46
  47 #include "kernelutil_x86_avx_128_fma_single.h"
  48
  49 /*
  50  * Gromacs nonbonded kernel:   nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_VF_avx_128_fma_single
  51  * Electrostatics interaction: Coulomb
  52  * VdW interaction:            CubicSplineTable
  53  * Geometry:                   Water3-Water3
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_VF_avx_128_fma_single
  58                     (t_nblist                    * gmx_restrict       nlist,
  59                      rvec                        * gmx_restrict          xx,
  60                      rvec                        * gmx_restrict          ff,
  61                      struct t_forcerec           * gmx_restrict          fr,
  62                      t_mdatoms                   * gmx_restrict     mdatoms,
  63                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  64                      t_nrnb                      * gmx_restrict        nrnb)
  65 {
  66     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  67      * just 0 for non-waters.
  68      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
  69      * jnr indices corresponding to data put in the four positions in the SIMD register.
  70      */
  71     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  72     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  73     int              jnrA,jnrB,jnrC,jnrD;
  74     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  75     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  76     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  77     real             rcutoff_scalar;
  78     real             *shiftvec,*fshift,*x,*f;
  79     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  80     real             scratch[4*DIM];
  81     __m128           fscal,rcutoff,rcutoff2,jidxall;
  82     int              vdwioffset0;
  83     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  84     int              vdwioffset1;
  85     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  86     int              vdwioffset2;
  87     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  88     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  89     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  90     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
  91     __m128           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  92     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
  93     __m128           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  94     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  95     __m128           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
  96     __m128           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
  97     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  98     __m128           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
  99     __m128           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 100     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 101     __m128           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
 102     __m128           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
 103     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 104     real             *charge;
 105     int              nvdwtype;
 106     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 107     int              *vdwtype;
 108     real             *vdwparam;
 109     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 110     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 111     __m128i          vfitab;
 112     __m128i          ifour       = _mm_set1_epi32(4);
 113     __m128           rt,vfeps,twovfeps,vftabscale,Y,F,G,H,Fp,VV,FF;
 114     real             *vftab;
 115     __m128           dummy_mask,cutoff_mask;
 116     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 117     __m128           one     = _mm_set1_ps(1.0);
 118     __m128           two     = _mm_set1_ps(2.0);
 119     x                = xx[0];
 120     f                = ff[0];
 121
 122     nri              = nlist->nri;
 123     iinr             = nlist->iinr;
 124     jindex           = nlist->jindex;
 125     jjnr             = nlist->jjnr;
 126     shiftidx         = nlist->shift;
 127     gid              = nlist->gid;
 128     shiftvec         = fr->shift_vec[0];
 129     fshift           = fr->fshift[0];
 130     facel            = _mm_set1_ps(fr->ic->epsfac);
 131     charge           = mdatoms->chargeA;
 132     nvdwtype         = fr->ntype;
 133     vdwparam         = fr->nbfp;
 134     vdwtype          = mdatoms->typeA;
 135
 136     vftab            = kernel_data->table_vdw->data;
 137     vftabscale       = _mm_set1_ps(kernel_data->table_vdw->scale);
 138
 139     /* Setup water-specific parameters */
 140     inr              = nlist->iinr[0];
 141     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
 142     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 143     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 144     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 145
 146     jq0              = _mm_set1_ps(charge[inr+0]);
 147     jq1              = _mm_set1_ps(charge[inr+1]);
 148     jq2              = _mm_set1_ps(charge[inr+2]);
 149     vdwjidx0A        = 2*vdwtype[inr+0];
 150     qq00             = _mm_mul_ps(iq0,jq0);
 151     c6_00            = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
 152     c12_00           = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
 153     qq01             = _mm_mul_ps(iq0,jq1);
 154     qq02             = _mm_mul_ps(iq0,jq2);
 155     qq10             = _mm_mul_ps(iq1,jq0);
 156     qq11             = _mm_mul_ps(iq1,jq1);
 157     qq12             = _mm_mul_ps(iq1,jq2);
 158     qq20             = _mm_mul_ps(iq2,jq0);
 159     qq21             = _mm_mul_ps(iq2,jq1);
 160     qq22             = _mm_mul_ps(iq2,jq2);
 161
 162     /* Avoid stupid compiler warnings */
 163     jnrA = jnrB = jnrC = jnrD = 0;
 164     j_coord_offsetA = 0;
 165     j_coord_offsetB = 0;
 166     j_coord_offsetC = 0;
 167     j_coord_offsetD = 0;
 168
 169     outeriter        = 0;
 170     inneriter        = 0;
 171
 172     for(iidx=0;iidx<4*DIM;iidx++)
 173     {
 174         scratch[iidx] = 0.0;
 175     }
 176
 177     /* Start outer loop over neighborlists */
 178     for(iidx=0; iidx<nri; iidx++)
 179     {
 180         /* Load shift vector for this list */
 181         i_shift_offset   = DIM*shiftidx[iidx];
 182
 183         /* Load limits for loop over neighbors */
 184         j_index_start    = jindex[iidx];
 185         j_index_end      = jindex[iidx+1];
 186
 187         /* Get outer coordinate index */
 188         inr              = iinr[iidx];
 189         i_coord_offset   = DIM*inr;
 190
 191         /* Load i particle coords and add shift vector */
 192         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 193                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 194
 195         fix0             = _mm_setzero_ps();
 196         fiy0             = _mm_setzero_ps();
 197         fiz0             = _mm_setzero_ps();
 198         fix1             = _mm_setzero_ps();
 199         fiy1             = _mm_setzero_ps();
 200         fiz1             = _mm_setzero_ps();
 201         fix2             = _mm_setzero_ps();
 202         fiy2             = _mm_setzero_ps();
 203         fiz2             = _mm_setzero_ps();
 204
 205         /* Reset potential sums */
 206         velecsum         = _mm_setzero_ps();
 207         vvdwsum          = _mm_setzero_ps();
 208
 209         /* Start inner kernel loop */
 210         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 211         {
 212
 213             /* Get j neighbor index, and coordinate index */
 214             jnrA             = jjnr[jidx];
 215             jnrB             = jjnr[jidx+1];
 216             jnrC             = jjnr[jidx+2];
 217             jnrD             = jjnr[jidx+3];
 218             j_coord_offsetA  = DIM*jnrA;
 219             j_coord_offsetB  = DIM*jnrB;
 220             j_coord_offsetC  = DIM*jnrC;
 221             j_coord_offsetD  = DIM*jnrD;
 222
 223             /* load j atom coordinates */
 224             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 225                                               x+j_coord_offsetC,x+j_coord_offsetD,
 226                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 227
 228             /* Calculate displacement vector */
 229             dx00             = _mm_sub_ps(ix0,jx0);
 230             dy00             = _mm_sub_ps(iy0,jy0);
 231             dz00             = _mm_sub_ps(iz0,jz0);
 232             dx01             = _mm_sub_ps(ix0,jx1);
 233             dy01             = _mm_sub_ps(iy0,jy1);
 234             dz01             = _mm_sub_ps(iz0,jz1);
 235             dx02             = _mm_sub_ps(ix0,jx2);
 236             dy02             = _mm_sub_ps(iy0,jy2);
 237             dz02             = _mm_sub_ps(iz0,jz2);
 238             dx10             = _mm_sub_ps(ix1,jx0);
 239             dy10             = _mm_sub_ps(iy1,jy0);
 240             dz10             = _mm_sub_ps(iz1,jz0);
 241             dx11             = _mm_sub_ps(ix1,jx1);
 242             dy11             = _mm_sub_ps(iy1,jy1);
 243             dz11             = _mm_sub_ps(iz1,jz1);
 244             dx12             = _mm_sub_ps(ix1,jx2);
 245             dy12             = _mm_sub_ps(iy1,jy2);
 246             dz12             = _mm_sub_ps(iz1,jz2);
 247             dx20             = _mm_sub_ps(ix2,jx0);
 248             dy20             = _mm_sub_ps(iy2,jy0);
 249             dz20             = _mm_sub_ps(iz2,jz0);
 250             dx21             = _mm_sub_ps(ix2,jx1);
 251             dy21             = _mm_sub_ps(iy2,jy1);
 252             dz21             = _mm_sub_ps(iz2,jz1);
 253             dx22             = _mm_sub_ps(ix2,jx2);
 254             dy22             = _mm_sub_ps(iy2,jy2);
 255             dz22             = _mm_sub_ps(iz2,jz2);
 256
 257             /* Calculate squared distance and things based on it */
 258             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 259             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
 260             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
 261             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 262             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
 263             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
 264             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 265             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
 266             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
 267
 268             rinv00           = avx128fma_invsqrt_f(rsq00);
 269             rinv01           = avx128fma_invsqrt_f(rsq01);
 270             rinv02           = avx128fma_invsqrt_f(rsq02);
 271             rinv10           = avx128fma_invsqrt_f(rsq10);
 272             rinv11           = avx128fma_invsqrt_f(rsq11);
 273             rinv12           = avx128fma_invsqrt_f(rsq12);
 274             rinv20           = avx128fma_invsqrt_f(rsq20);
 275             rinv21           = avx128fma_invsqrt_f(rsq21);
 276             rinv22           = avx128fma_invsqrt_f(rsq22);
 277
 278             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 279             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
 280             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
 281             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 282             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
 283             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
 284             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 285             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
 286             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
 287
 288             fjx0             = _mm_setzero_ps();
 289             fjy0             = _mm_setzero_ps();
 290             fjz0             = _mm_setzero_ps();
 291             fjx1             = _mm_setzero_ps();
 292             fjy1             = _mm_setzero_ps();
 293             fjz1             = _mm_setzero_ps();
 294             fjx2             = _mm_setzero_ps();
 295             fjy2             = _mm_setzero_ps();
 296             fjz2             = _mm_setzero_ps();
 297
 298             /**************************
 299              * CALCULATE INTERACTIONS *
 300              **************************/
 301
 302             r00              = _mm_mul_ps(rsq00,rinv00);
 303
 304             /* Calculate table index by multiplying r with table scale and truncate to integer */
 305             rt               = _mm_mul_ps(r00,vftabscale);
 306             vfitab           = _mm_cvttps_epi32(rt);
 307 #ifdef __XOP__
 308             vfeps            = _mm_frcz_ps(rt);
 309 #else
 310             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
 311 #endif
 312             twovfeps         = _mm_add_ps(vfeps,vfeps);
 313             vfitab           = _mm_slli_epi32(vfitab,3);
 314
 315             /* COULOMB ELECTROSTATICS */
 316             velec            = _mm_mul_ps(qq00,rinv00);
 317             felec            = _mm_mul_ps(velec,rinvsq00);
 318
 319             /* CUBIC SPLINE TABLE DISPERSION */
 320             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 321             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 322             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 323             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 324             _MM_TRANSPOSE4_PS(Y,F,G,H);
 325             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 326             VV               = _mm_macc_ps(vfeps,Fp,Y);
 327             vvdw6            = _mm_mul_ps(c6_00,VV);
 328             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 329             fvdw6            = _mm_mul_ps(c6_00,FF);
 330
 331             /* CUBIC SPLINE TABLE REPULSION */
 332             vfitab           = _mm_add_epi32(vfitab,ifour);
 333             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 334             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 335             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 336             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 337             _MM_TRANSPOSE4_PS(Y,F,G,H);
 338             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 339             VV               = _mm_macc_ps(vfeps,Fp,Y);
 340             vvdw12           = _mm_mul_ps(c12_00,VV);
 341             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 342             fvdw12           = _mm_mul_ps(c12_00,FF);
 343             vvdw             = _mm_add_ps(vvdw12,vvdw6);
 344             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
 345
 346             /* Update potential sum for this i atom from the interaction with this j atom. */
 347             velecsum         = _mm_add_ps(velecsum,velec);
 348             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 349
 350             fscal            = _mm_add_ps(felec,fvdw);
 351
 352              /* Update vectorial force */
 353             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 354             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 355             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 356
 357             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 358             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 359             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 360
 361             /**************************
 362              * CALCULATE INTERACTIONS *
 363              **************************/
 364
 365             /* COULOMB ELECTROSTATICS */
 366             velec            = _mm_mul_ps(qq01,rinv01);
 367             felec            = _mm_mul_ps(velec,rinvsq01);
 368
 369             /* Update potential sum for this i atom from the interaction with this j atom. */
 370             velecsum         = _mm_add_ps(velecsum,velec);
 371
 372             fscal            = felec;
 373
 374              /* Update vectorial force */
 375             fix0             = _mm_macc_ps(dx01,fscal,fix0);
 376             fiy0             = _mm_macc_ps(dy01,fscal,fiy0);
 377             fiz0             = _mm_macc_ps(dz01,fscal,fiz0);
 378
 379             fjx1             = _mm_macc_ps(dx01,fscal,fjx1);
 380             fjy1             = _mm_macc_ps(dy01,fscal,fjy1);
 381             fjz1             = _mm_macc_ps(dz01,fscal,fjz1);
 382
 383             /**************************
 384              * CALCULATE INTERACTIONS *
 385              **************************/
 386
 387             /* COULOMB ELECTROSTATICS */
 388             velec            = _mm_mul_ps(qq02,rinv02);
 389             felec            = _mm_mul_ps(velec,rinvsq02);
 390
 391             /* Update potential sum for this i atom from the interaction with this j atom. */
 392             velecsum         = _mm_add_ps(velecsum,velec);
 393
 394             fscal            = felec;
 395
 396              /* Update vectorial force */
 397             fix0             = _mm_macc_ps(dx02,fscal,fix0);
 398             fiy0             = _mm_macc_ps(dy02,fscal,fiy0);
 399             fiz0             = _mm_macc_ps(dz02,fscal,fiz0);
 400
 401             fjx2             = _mm_macc_ps(dx02,fscal,fjx2);
 402             fjy2             = _mm_macc_ps(dy02,fscal,fjy2);
 403             fjz2             = _mm_macc_ps(dz02,fscal,fjz2);
 404
 405             /**************************
 406              * CALCULATE INTERACTIONS *
 407              **************************/
 408
 409             /* COULOMB ELECTROSTATICS */
 410             velec            = _mm_mul_ps(qq10,rinv10);
 411             felec            = _mm_mul_ps(velec,rinvsq10);
 412
 413             /* Update potential sum for this i atom from the interaction with this j atom. */
 414             velecsum         = _mm_add_ps(velecsum,velec);
 415
 416             fscal            = felec;
 417
 418              /* Update vectorial force */
 419             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 420             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 421             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 422
 423             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 424             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 425             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 426
 427             /**************************
 428              * CALCULATE INTERACTIONS *
 429              **************************/
 430
 431             /* COULOMB ELECTROSTATICS */
 432             velec            = _mm_mul_ps(qq11,rinv11);
 433             felec            = _mm_mul_ps(velec,rinvsq11);
 434
 435             /* Update potential sum for this i atom from the interaction with this j atom. */
 436             velecsum         = _mm_add_ps(velecsum,velec);
 437
 438             fscal            = felec;
 439
 440              /* Update vectorial force */
 441             fix1             = _mm_macc_ps(dx11,fscal,fix1);
 442             fiy1             = _mm_macc_ps(dy11,fscal,fiy1);
 443             fiz1             = _mm_macc_ps(dz11,fscal,fiz1);
 444
 445             fjx1             = _mm_macc_ps(dx11,fscal,fjx1);
 446             fjy1             = _mm_macc_ps(dy11,fscal,fjy1);
 447             fjz1             = _mm_macc_ps(dz11,fscal,fjz1);
 448
 449             /**************************
 450              * CALCULATE INTERACTIONS *
 451              **************************/
 452
 453             /* COULOMB ELECTROSTATICS */
 454             velec            = _mm_mul_ps(qq12,rinv12);
 455             felec            = _mm_mul_ps(velec,rinvsq12);
 456
 457             /* Update potential sum for this i atom from the interaction with this j atom. */
 458             velecsum         = _mm_add_ps(velecsum,velec);
 459
 460             fscal            = felec;
 461
 462              /* Update vectorial force */
 463             fix1             = _mm_macc_ps(dx12,fscal,fix1);
 464             fiy1             = _mm_macc_ps(dy12,fscal,fiy1);
 465             fiz1             = _mm_macc_ps(dz12,fscal,fiz1);
 466
 467             fjx2             = _mm_macc_ps(dx12,fscal,fjx2);
 468             fjy2             = _mm_macc_ps(dy12,fscal,fjy2);
 469             fjz2             = _mm_macc_ps(dz12,fscal,fjz2);
 470
 471             /**************************
 472              * CALCULATE INTERACTIONS *
 473              **************************/
 474
 475             /* COULOMB ELECTROSTATICS */
 476             velec            = _mm_mul_ps(qq20,rinv20);
 477             felec            = _mm_mul_ps(velec,rinvsq20);
 478
 479             /* Update potential sum for this i atom from the interaction with this j atom. */
 480             velecsum         = _mm_add_ps(velecsum,velec);
 481
 482             fscal            = felec;
 483
 484              /* Update vectorial force */
 485             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 486             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 487             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 488
 489             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 490             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 491             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 492
 493             /**************************
 494              * CALCULATE INTERACTIONS *
 495              **************************/
 496
 497             /* COULOMB ELECTROSTATICS */
 498             velec            = _mm_mul_ps(qq21,rinv21);
 499             felec            = _mm_mul_ps(velec,rinvsq21);
 500
 501             /* Update potential sum for this i atom from the interaction with this j atom. */
 502             velecsum         = _mm_add_ps(velecsum,velec);
 503
 504             fscal            = felec;
 505
 506              /* Update vectorial force */
 507             fix2             = _mm_macc_ps(dx21,fscal,fix2);
 508             fiy2             = _mm_macc_ps(dy21,fscal,fiy2);
 509             fiz2             = _mm_macc_ps(dz21,fscal,fiz2);
 510
 511             fjx1             = _mm_macc_ps(dx21,fscal,fjx1);
 512             fjy1             = _mm_macc_ps(dy21,fscal,fjy1);
 513             fjz1             = _mm_macc_ps(dz21,fscal,fjz1);
 514
 515             /**************************
 516              * CALCULATE INTERACTIONS *
 517              **************************/
 518
 519             /* COULOMB ELECTROSTATICS */
 520             velec            = _mm_mul_ps(qq22,rinv22);
 521             felec            = _mm_mul_ps(velec,rinvsq22);
 522
 523             /* Update potential sum for this i atom from the interaction with this j atom. */
 524             velecsum         = _mm_add_ps(velecsum,velec);
 525
 526             fscal            = felec;
 527
 528              /* Update vectorial force */
 529             fix2             = _mm_macc_ps(dx22,fscal,fix2);
 530             fiy2             = _mm_macc_ps(dy22,fscal,fiy2);
 531             fiz2             = _mm_macc_ps(dz22,fscal,fiz2);
 532
 533             fjx2             = _mm_macc_ps(dx22,fscal,fjx2);
 534             fjy2             = _mm_macc_ps(dy22,fscal,fjy2);
 535             fjz2             = _mm_macc_ps(dz22,fscal,fjz2);
 536
 537             fjptrA             = f+j_coord_offsetA;
 538             fjptrB             = f+j_coord_offsetB;
 539             fjptrC             = f+j_coord_offsetC;
 540             fjptrD             = f+j_coord_offsetD;
 541
 542             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 543                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 544
 545             /* Inner loop uses 314 flops */
 546         }
 547
 548         if(jidx<j_index_end)
 549         {
 550
 551             /* Get j neighbor index, and coordinate index */
 552             jnrlistA         = jjnr[jidx];
 553             jnrlistB         = jjnr[jidx+1];
 554             jnrlistC         = jjnr[jidx+2];
 555             jnrlistD         = jjnr[jidx+3];
 556             /* Sign of each element will be negative for non-real atoms.
 557              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 558              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 559              */
 560             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 561             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 562             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 563             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 564             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 565             j_coord_offsetA  = DIM*jnrA;
 566             j_coord_offsetB  = DIM*jnrB;
 567             j_coord_offsetC  = DIM*jnrC;
 568             j_coord_offsetD  = DIM*jnrD;
 569
 570             /* load j atom coordinates */
 571             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 572                                               x+j_coord_offsetC,x+j_coord_offsetD,
 573                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
 574
 575             /* Calculate displacement vector */
 576             dx00             = _mm_sub_ps(ix0,jx0);
 577             dy00             = _mm_sub_ps(iy0,jy0);
 578             dz00             = _mm_sub_ps(iz0,jz0);
 579             dx01             = _mm_sub_ps(ix0,jx1);
 580             dy01             = _mm_sub_ps(iy0,jy1);
 581             dz01             = _mm_sub_ps(iz0,jz1);
 582             dx02             = _mm_sub_ps(ix0,jx2);
 583             dy02             = _mm_sub_ps(iy0,jy2);
 584             dz02             = _mm_sub_ps(iz0,jz2);
 585             dx10             = _mm_sub_ps(ix1,jx0);
 586             dy10             = _mm_sub_ps(iy1,jy0);
 587             dz10             = _mm_sub_ps(iz1,jz0);
 588             dx11             = _mm_sub_ps(ix1,jx1);
 589             dy11             = _mm_sub_ps(iy1,jy1);
 590             dz11             = _mm_sub_ps(iz1,jz1);
 591             dx12             = _mm_sub_ps(ix1,jx2);
 592             dy12             = _mm_sub_ps(iy1,jy2);
 593             dz12             = _mm_sub_ps(iz1,jz2);
 594             dx20             = _mm_sub_ps(ix2,jx0);
 595             dy20             = _mm_sub_ps(iy2,jy0);
 596             dz20             = _mm_sub_ps(iz2,jz0);
 597             dx21             = _mm_sub_ps(ix2,jx1);
 598             dy21             = _mm_sub_ps(iy2,jy1);
 599             dz21             = _mm_sub_ps(iz2,jz1);
 600             dx22             = _mm_sub_ps(ix2,jx2);
 601             dy22             = _mm_sub_ps(iy2,jy2);
 602             dz22             = _mm_sub_ps(iz2,jz2);
 603
 604             /* Calculate squared distance and things based on it */
 605             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 606             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
 607             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
 608             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 609             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
 610             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
 611             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 612             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
 613             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
 614
 615             rinv00           = avx128fma_invsqrt_f(rsq00);
 616             rinv01           = avx128fma_invsqrt_f(rsq01);
 617             rinv02           = avx128fma_invsqrt_f(rsq02);
 618             rinv10           = avx128fma_invsqrt_f(rsq10);
 619             rinv11           = avx128fma_invsqrt_f(rsq11);
 620             rinv12           = avx128fma_invsqrt_f(rsq12);
 621             rinv20           = avx128fma_invsqrt_f(rsq20);
 622             rinv21           = avx128fma_invsqrt_f(rsq21);
 623             rinv22           = avx128fma_invsqrt_f(rsq22);
 624
 625             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 626             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
 627             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
 628             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 629             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
 630             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
 631             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 632             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
 633             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
 634
 635             fjx0             = _mm_setzero_ps();
 636             fjy0             = _mm_setzero_ps();
 637             fjz0             = _mm_setzero_ps();
 638             fjx1             = _mm_setzero_ps();
 639             fjy1             = _mm_setzero_ps();
 640             fjz1             = _mm_setzero_ps();
 641             fjx2             = _mm_setzero_ps();
 642             fjy2             = _mm_setzero_ps();
 643             fjz2             = _mm_setzero_ps();
 644
 645             /**************************
 646              * CALCULATE INTERACTIONS *
 647              **************************/
 648
 649             r00              = _mm_mul_ps(rsq00,rinv00);
 650             r00              = _mm_andnot_ps(dummy_mask,r00);
 651
 652             /* Calculate table index by multiplying r with table scale and truncate to integer */
 653             rt               = _mm_mul_ps(r00,vftabscale);
 654             vfitab           = _mm_cvttps_epi32(rt);
 655 #ifdef __XOP__
 656             vfeps            = _mm_frcz_ps(rt);
 657 #else
 658             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
 659 #endif
 660             twovfeps         = _mm_add_ps(vfeps,vfeps);
 661             vfitab           = _mm_slli_epi32(vfitab,3);
 662
 663             /* COULOMB ELECTROSTATICS */
 664             velec            = _mm_mul_ps(qq00,rinv00);
 665             felec            = _mm_mul_ps(velec,rinvsq00);
 666
 667             /* CUBIC SPLINE TABLE DISPERSION */
 668             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 669             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 670             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 671             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 672             _MM_TRANSPOSE4_PS(Y,F,G,H);
 673             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 674             VV               = _mm_macc_ps(vfeps,Fp,Y);
 675             vvdw6            = _mm_mul_ps(c6_00,VV);
 676             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 677             fvdw6            = _mm_mul_ps(c6_00,FF);
 678
 679             /* CUBIC SPLINE TABLE REPULSION */
 680             vfitab           = _mm_add_epi32(vfitab,ifour);
 681             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 682             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 683             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 684             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 685             _MM_TRANSPOSE4_PS(Y,F,G,H);
 686             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 687             VV               = _mm_macc_ps(vfeps,Fp,Y);
 688             vvdw12           = _mm_mul_ps(c12_00,VV);
 689             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 690             fvdw12           = _mm_mul_ps(c12_00,FF);
 691             vvdw             = _mm_add_ps(vvdw12,vvdw6);
 692             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
 693
 694             /* Update potential sum for this i atom from the interaction with this j atom. */
 695             velec            = _mm_andnot_ps(dummy_mask,velec);
 696             velecsum         = _mm_add_ps(velecsum,velec);
 697             vvdw             = _mm_andnot_ps(dummy_mask,vvdw);
 698             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 699
 700             fscal            = _mm_add_ps(felec,fvdw);
 701
 702             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 703
 704              /* Update vectorial force */
 705             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 706             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 707             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 708
 709             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 710             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 711             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 712
 713             /**************************
 714              * CALCULATE INTERACTIONS *
 715              **************************/
 716
 717             /* COULOMB ELECTROSTATICS */
 718             velec            = _mm_mul_ps(qq01,rinv01);
 719             felec            = _mm_mul_ps(velec,rinvsq01);
 720
 721             /* Update potential sum for this i atom from the interaction with this j atom. */
 722             velec            = _mm_andnot_ps(dummy_mask,velec);
 723             velecsum         = _mm_add_ps(velecsum,velec);
 724
 725             fscal            = felec;
 726
 727             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 728
 729              /* Update vectorial force */
 730             fix0             = _mm_macc_ps(dx01,fscal,fix0);
 731             fiy0             = _mm_macc_ps(dy01,fscal,fiy0);
 732             fiz0             = _mm_macc_ps(dz01,fscal,fiz0);
 733
 734             fjx1             = _mm_macc_ps(dx01,fscal,fjx1);
 735             fjy1             = _mm_macc_ps(dy01,fscal,fjy1);
 736             fjz1             = _mm_macc_ps(dz01,fscal,fjz1);
 737
 738             /**************************
 739              * CALCULATE INTERACTIONS *
 740              **************************/
 741
 742             /* COULOMB ELECTROSTATICS */
 743             velec            = _mm_mul_ps(qq02,rinv02);
 744             felec            = _mm_mul_ps(velec,rinvsq02);
 745
 746             /* Update potential sum for this i atom from the interaction with this j atom. */
 747             velec            = _mm_andnot_ps(dummy_mask,velec);
 748             velecsum         = _mm_add_ps(velecsum,velec);
 749
 750             fscal            = felec;
 751
 752             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 753
 754              /* Update vectorial force */
 755             fix0             = _mm_macc_ps(dx02,fscal,fix0);
 756             fiy0             = _mm_macc_ps(dy02,fscal,fiy0);
 757             fiz0             = _mm_macc_ps(dz02,fscal,fiz0);
 758
 759             fjx2             = _mm_macc_ps(dx02,fscal,fjx2);
 760             fjy2             = _mm_macc_ps(dy02,fscal,fjy2);
 761             fjz2             = _mm_macc_ps(dz02,fscal,fjz2);
 762
 763             /**************************
 764              * CALCULATE INTERACTIONS *
 765              **************************/
 766
 767             /* COULOMB ELECTROSTATICS */
 768             velec            = _mm_mul_ps(qq10,rinv10);
 769             felec            = _mm_mul_ps(velec,rinvsq10);
 770
 771             /* Update potential sum for this i atom from the interaction with this j atom. */
 772             velec            = _mm_andnot_ps(dummy_mask,velec);
 773             velecsum         = _mm_add_ps(velecsum,velec);
 774
 775             fscal            = felec;
 776
 777             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 778
 779              /* Update vectorial force */
 780             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 781             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 782             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 783
 784             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 785             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 786             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 787
 788             /**************************
 789              * CALCULATE INTERACTIONS *
 790              **************************/
 791
 792             /* COULOMB ELECTROSTATICS */
 793             velec            = _mm_mul_ps(qq11,rinv11);
 794             felec            = _mm_mul_ps(velec,rinvsq11);
 795
 796             /* Update potential sum for this i atom from the interaction with this j atom. */
 797             velec            = _mm_andnot_ps(dummy_mask,velec);
 798             velecsum         = _mm_add_ps(velecsum,velec);
 799
 800             fscal            = felec;
 801
 802             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 803
 804              /* Update vectorial force */
 805             fix1             = _mm_macc_ps(dx11,fscal,fix1);
 806             fiy1             = _mm_macc_ps(dy11,fscal,fiy1);
 807             fiz1             = _mm_macc_ps(dz11,fscal,fiz1);
 808
 809             fjx1             = _mm_macc_ps(dx11,fscal,fjx1);
 810             fjy1             = _mm_macc_ps(dy11,fscal,fjy1);
 811             fjz1             = _mm_macc_ps(dz11,fscal,fjz1);
 812
 813             /**************************
 814              * CALCULATE INTERACTIONS *
 815              **************************/
 816
 817             /* COULOMB ELECTROSTATICS */
 818             velec            = _mm_mul_ps(qq12,rinv12);
 819             felec            = _mm_mul_ps(velec,rinvsq12);
 820
 821             /* Update potential sum for this i atom from the interaction with this j atom. */
 822             velec            = _mm_andnot_ps(dummy_mask,velec);
 823             velecsum         = _mm_add_ps(velecsum,velec);
 824
 825             fscal            = felec;
 826
 827             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 828
 829              /* Update vectorial force */
 830             fix1             = _mm_macc_ps(dx12,fscal,fix1);
 831             fiy1             = _mm_macc_ps(dy12,fscal,fiy1);
 832             fiz1             = _mm_macc_ps(dz12,fscal,fiz1);
 833
 834             fjx2             = _mm_macc_ps(dx12,fscal,fjx2);
 835             fjy2             = _mm_macc_ps(dy12,fscal,fjy2);
 836             fjz2             = _mm_macc_ps(dz12,fscal,fjz2);
 837
 838             /**************************
 839              * CALCULATE INTERACTIONS *
 840              **************************/
 841
 842             /* COULOMB ELECTROSTATICS */
 843             velec            = _mm_mul_ps(qq20,rinv20);
 844             felec            = _mm_mul_ps(velec,rinvsq20);
 845
 846             /* Update potential sum for this i atom from the interaction with this j atom. */
 847             velec            = _mm_andnot_ps(dummy_mask,velec);
 848             velecsum         = _mm_add_ps(velecsum,velec);
 849
 850             fscal            = felec;
 851
 852             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 853
 854              /* Update vectorial force */
 855             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 856             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 857             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 858
 859             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 860             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 861             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 862
 863             /**************************
 864              * CALCULATE INTERACTIONS *
 865              **************************/
 866
 867             /* COULOMB ELECTROSTATICS */
 868             velec            = _mm_mul_ps(qq21,rinv21);
 869             felec            = _mm_mul_ps(velec,rinvsq21);
 870
 871             /* Update potential sum for this i atom from the interaction with this j atom. */
 872             velec            = _mm_andnot_ps(dummy_mask,velec);
 873             velecsum         = _mm_add_ps(velecsum,velec);
 874
 875             fscal            = felec;
 876
 877             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 878
 879              /* Update vectorial force */
 880             fix2             = _mm_macc_ps(dx21,fscal,fix2);
 881             fiy2             = _mm_macc_ps(dy21,fscal,fiy2);
 882             fiz2             = _mm_macc_ps(dz21,fscal,fiz2);
 883
 884             fjx1             = _mm_macc_ps(dx21,fscal,fjx1);
 885             fjy1             = _mm_macc_ps(dy21,fscal,fjy1);
 886             fjz1             = _mm_macc_ps(dz21,fscal,fjz1);
 887
 888             /**************************
 889              * CALCULATE INTERACTIONS *
 890              **************************/
 891
 892             /* COULOMB ELECTROSTATICS */
 893             velec            = _mm_mul_ps(qq22,rinv22);
 894             felec            = _mm_mul_ps(velec,rinvsq22);
 895
 896             /* Update potential sum for this i atom from the interaction with this j atom. */
 897             velec            = _mm_andnot_ps(dummy_mask,velec);
 898             velecsum         = _mm_add_ps(velecsum,velec);
 899
 900             fscal            = felec;
 901
 902             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 903
 904              /* Update vectorial force */
 905             fix2             = _mm_macc_ps(dx22,fscal,fix2);
 906             fiy2             = _mm_macc_ps(dy22,fscal,fiy2);
 907             fiz2             = _mm_macc_ps(dz22,fscal,fiz2);
 908
 909             fjx2             = _mm_macc_ps(dx22,fscal,fjx2);
 910             fjy2             = _mm_macc_ps(dy22,fscal,fjy2);
 911             fjz2             = _mm_macc_ps(dz22,fscal,fjz2);
 912
 913             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 914             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 915             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 916             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 917
 918             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 919                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
 920
 921             /* Inner loop uses 315 flops */
 922         }
 923
 924         /* End of innermost loop */
 925
 926         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 927                                               f+i_coord_offset,fshift+i_shift_offset);
 928
 929         ggid                        = gid[iidx];
 930         /* Update potential energies */
 931         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 932         gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 933
 934         /* Increment number of inner iterations */
 935         inneriter                  += j_index_end - j_index_start;
 936
 937         /* Outer loop uses 20 flops */
 938     }
 939
 940     /* Increment number of outer iterations */
 941     outeriter        += nri;
 942
 943     /* Update outer/inner flops */
 944
 945     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_VF,outeriter*20 + inneriter*315);
 946 }
 947 /*
 948  * Gromacs nonbonded kernel:   nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_avx_128_fma_single
 949  * Electrostatics interaction: Coulomb
 950  * VdW interaction:            CubicSplineTable
 951  * Geometry:                   Water3-Water3
 952  * Calculate force/pot:        Force
 953  */
 954 void
 955 nb_kernel_ElecCoul_VdwCSTab_GeomW3W3_F_avx_128_fma_single
 956                     (t_nblist                    * gmx_restrict       nlist,
 957                      rvec                        * gmx_restrict          xx,
 958                      rvec                        * gmx_restrict          ff,
 959                      struct t_forcerec           * gmx_restrict          fr,
 960                      t_mdatoms                   * gmx_restrict     mdatoms,
 961                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 962                      t_nrnb                      * gmx_restrict        nrnb)
 963 {
 964     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 965      * just 0 for non-waters.
 966      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
 967      * jnr indices corresponding to data put in the four positions in the SIMD register.
 968      */
 969     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 970     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 971     int              jnrA,jnrB,jnrC,jnrD;
 972     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 973     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 974     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 975     real             rcutoff_scalar;
 976     real             *shiftvec,*fshift,*x,*f;
 977     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 978     real             scratch[4*DIM];
 979     __m128           fscal,rcutoff,rcutoff2,jidxall;
 980     int              vdwioffset0;
 981     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 982     int              vdwioffset1;
 983     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 984     int              vdwioffset2;
 985     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 986     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 987     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 988     int              vdwjidx1A,vdwjidx1B,vdwjidx1C,vdwjidx1D;
 989     __m128           jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 990     int              vdwjidx2A,vdwjidx2B,vdwjidx2C,vdwjidx2D;
 991     __m128           jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 992     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 993     __m128           dx01,dy01,dz01,rsq01,rinv01,rinvsq01,r01,qq01,c6_01,c12_01;
 994     __m128           dx02,dy02,dz02,rsq02,rinv02,rinvsq02,r02,qq02,c6_02,c12_02;
 995     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 996     __m128           dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11;
 997     __m128           dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12;
 998     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 999     __m128           dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21;
1000     __m128           dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22;
1001     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
1002     real             *charge;
1003     int              nvdwtype;
1004     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
1005     int              *vdwtype;
1006     real             *vdwparam;
1007     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
1008     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
1009     __m128i          vfitab;
1010     __m128i          ifour       = _mm_set1_epi32(4);
1011     __m128           rt,vfeps,twovfeps,vftabscale,Y,F,G,H,Fp,VV,FF;
1012     real             *vftab;
1013     __m128           dummy_mask,cutoff_mask;
1014     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
1015     __m128           one     = _mm_set1_ps(1.0);
1016     __m128           two     = _mm_set1_ps(2.0);
1017     x                = xx[0];
1018     f                = ff[0];
1019
1020     nri              = nlist->nri;
1021     iinr             = nlist->iinr;
1022     jindex           = nlist->jindex;
1023     jjnr             = nlist->jjnr;
1024     shiftidx         = nlist->shift;
1025     gid              = nlist->gid;
1026     shiftvec         = fr->shift_vec[0];
1027     fshift           = fr->fshift[0];
1028     facel            = _mm_set1_ps(fr->ic->epsfac);
1029     charge           = mdatoms->chargeA;
1030     nvdwtype         = fr->ntype;
1031     vdwparam         = fr->nbfp;
1032     vdwtype          = mdatoms->typeA;
1033
1034     vftab            = kernel_data->table_vdw->data;
1035     vftabscale       = _mm_set1_ps(kernel_data->table_vdw->scale);
1036
1037     /* Setup water-specific parameters */
1038     inr              = nlist->iinr[0];
1039     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
1040     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
1041     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
1042     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
1043
1044     jq0              = _mm_set1_ps(charge[inr+0]);
1045     jq1              = _mm_set1_ps(charge[inr+1]);
1046     jq2              = _mm_set1_ps(charge[inr+2]);
1047     vdwjidx0A        = 2*vdwtype[inr+0];
1048     qq00             = _mm_mul_ps(iq0,jq0);
1049     c6_00            = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A]);
1050     c12_00           = _mm_set1_ps(vdwparam[vdwioffset0+vdwjidx0A+1]);
1051     qq01             = _mm_mul_ps(iq0,jq1);
1052     qq02             = _mm_mul_ps(iq0,jq2);
1053     qq10             = _mm_mul_ps(iq1,jq0);
1054     qq11             = _mm_mul_ps(iq1,jq1);
1055     qq12             = _mm_mul_ps(iq1,jq2);
1056     qq20             = _mm_mul_ps(iq2,jq0);
1057     qq21             = _mm_mul_ps(iq2,jq1);
1058     qq22             = _mm_mul_ps(iq2,jq2);
1059
1060     /* Avoid stupid compiler warnings */
1061     jnrA = jnrB = jnrC = jnrD = 0;
1062     j_coord_offsetA = 0;
1063     j_coord_offsetB = 0;
1064     j_coord_offsetC = 0;
1065     j_coord_offsetD = 0;
1066
1067     outeriter        = 0;
1068     inneriter        = 0;
1069
1070     for(iidx=0;iidx<4*DIM;iidx++)
1071     {
1072         scratch[iidx] = 0.0;
1073     }
1074
1075     /* Start outer loop over neighborlists */
1076     for(iidx=0; iidx<nri; iidx++)
1077     {
1078         /* Load shift vector for this list */
1079         i_shift_offset   = DIM*shiftidx[iidx];
1080
1081         /* Load limits for loop over neighbors */
1082         j_index_start    = jindex[iidx];
1083         j_index_end      = jindex[iidx+1];
1084
1085         /* Get outer coordinate index */
1086         inr              = iinr[iidx];
1087         i_coord_offset   = DIM*inr;
1088
1089         /* Load i particle coords and add shift vector */
1090         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
1091                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
1092
1093         fix0             = _mm_setzero_ps();
1094         fiy0             = _mm_setzero_ps();
1095         fiz0             = _mm_setzero_ps();
1096         fix1             = _mm_setzero_ps();
1097         fiy1             = _mm_setzero_ps();
1098         fiz1             = _mm_setzero_ps();
1099         fix2             = _mm_setzero_ps();
1100         fiy2             = _mm_setzero_ps();
1101         fiz2             = _mm_setzero_ps();
1102
1103         /* Start inner kernel loop */
1104         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
1105         {
1106
1107             /* Get j neighbor index, and coordinate index */
1108             jnrA             = jjnr[jidx];
1109             jnrB             = jjnr[jidx+1];
1110             jnrC             = jjnr[jidx+2];
1111             jnrD             = jjnr[jidx+3];
1112             j_coord_offsetA  = DIM*jnrA;
1113             j_coord_offsetB  = DIM*jnrB;
1114             j_coord_offsetC  = DIM*jnrC;
1115             j_coord_offsetD  = DIM*jnrD;
1116
1117             /* load j atom coordinates */
1118             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1119                                               x+j_coord_offsetC,x+j_coord_offsetD,
1120                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1121
1122             /* Calculate displacement vector */
1123             dx00             = _mm_sub_ps(ix0,jx0);
1124             dy00             = _mm_sub_ps(iy0,jy0);
1125             dz00             = _mm_sub_ps(iz0,jz0);
1126             dx01             = _mm_sub_ps(ix0,jx1);
1127             dy01             = _mm_sub_ps(iy0,jy1);
1128             dz01             = _mm_sub_ps(iz0,jz1);
1129             dx02             = _mm_sub_ps(ix0,jx2);
1130             dy02             = _mm_sub_ps(iy0,jy2);
1131             dz02             = _mm_sub_ps(iz0,jz2);
1132             dx10             = _mm_sub_ps(ix1,jx0);
1133             dy10             = _mm_sub_ps(iy1,jy0);
1134             dz10             = _mm_sub_ps(iz1,jz0);
1135             dx11             = _mm_sub_ps(ix1,jx1);
1136             dy11             = _mm_sub_ps(iy1,jy1);
1137             dz11             = _mm_sub_ps(iz1,jz1);
1138             dx12             = _mm_sub_ps(ix1,jx2);
1139             dy12             = _mm_sub_ps(iy1,jy2);
1140             dz12             = _mm_sub_ps(iz1,jz2);
1141             dx20             = _mm_sub_ps(ix2,jx0);
1142             dy20             = _mm_sub_ps(iy2,jy0);
1143             dz20             = _mm_sub_ps(iz2,jz0);
1144             dx21             = _mm_sub_ps(ix2,jx1);
1145             dy21             = _mm_sub_ps(iy2,jy1);
1146             dz21             = _mm_sub_ps(iz2,jz1);
1147             dx22             = _mm_sub_ps(ix2,jx2);
1148             dy22             = _mm_sub_ps(iy2,jy2);
1149             dz22             = _mm_sub_ps(iz2,jz2);
1150
1151             /* Calculate squared distance and things based on it */
1152             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1153             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1154             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1155             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1156             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1157             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1158             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1159             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1160             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1161
1162             rinv00           = avx128fma_invsqrt_f(rsq00);
1163             rinv01           = avx128fma_invsqrt_f(rsq01);
1164             rinv02           = avx128fma_invsqrt_f(rsq02);
1165             rinv10           = avx128fma_invsqrt_f(rsq10);
1166             rinv11           = avx128fma_invsqrt_f(rsq11);
1167             rinv12           = avx128fma_invsqrt_f(rsq12);
1168             rinv20           = avx128fma_invsqrt_f(rsq20);
1169             rinv21           = avx128fma_invsqrt_f(rsq21);
1170             rinv22           = avx128fma_invsqrt_f(rsq22);
1171
1172             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1173             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
1174             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
1175             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1176             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
1177             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
1178             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1179             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
1180             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
1181
1182             fjx0             = _mm_setzero_ps();
1183             fjy0             = _mm_setzero_ps();
1184             fjz0             = _mm_setzero_ps();
1185             fjx1             = _mm_setzero_ps();
1186             fjy1             = _mm_setzero_ps();
1187             fjz1             = _mm_setzero_ps();
1188             fjx2             = _mm_setzero_ps();
1189             fjy2             = _mm_setzero_ps();
1190             fjz2             = _mm_setzero_ps();
1191
1192             /**************************
1193              * CALCULATE INTERACTIONS *
1194              **************************/
1195
1196             r00              = _mm_mul_ps(rsq00,rinv00);
1197
1198             /* Calculate table index by multiplying r with table scale and truncate to integer */
1199             rt               = _mm_mul_ps(r00,vftabscale);
1200             vfitab           = _mm_cvttps_epi32(rt);
1201 #ifdef __XOP__
1202             vfeps            = _mm_frcz_ps(rt);
1203 #else
1204             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
1205 #endif
1206             twovfeps         = _mm_add_ps(vfeps,vfeps);
1207             vfitab           = _mm_slli_epi32(vfitab,3);
1208
1209             /* COULOMB ELECTROSTATICS */
1210             velec            = _mm_mul_ps(qq00,rinv00);
1211             felec            = _mm_mul_ps(velec,rinvsq00);
1212
1213             /* CUBIC SPLINE TABLE DISPERSION */
1214             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1215             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1216             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1217             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1218             _MM_TRANSPOSE4_PS(Y,F,G,H);
1219             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1220             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1221             fvdw6            = _mm_mul_ps(c6_00,FF);
1222
1223             /* CUBIC SPLINE TABLE REPULSION */
1224             vfitab           = _mm_add_epi32(vfitab,ifour);
1225             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1226             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1227             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1228             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1229             _MM_TRANSPOSE4_PS(Y,F,G,H);
1230             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1231             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1232             fvdw12           = _mm_mul_ps(c12_00,FF);
1233             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
1234
1235             fscal            = _mm_add_ps(felec,fvdw);
1236
1237              /* Update vectorial force */
1238             fix0             = _mm_macc_ps(dx00,fscal,fix0);
1239             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
1240             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
1241
1242             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
1243             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
1244             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
1245
1246             /**************************
1247              * CALCULATE INTERACTIONS *
1248              **************************/
1249
1250             /* COULOMB ELECTROSTATICS */
1251             velec            = _mm_mul_ps(qq01,rinv01);
1252             felec            = _mm_mul_ps(velec,rinvsq01);
1253
1254             fscal            = felec;
1255
1256              /* Update vectorial force */
1257             fix0             = _mm_macc_ps(dx01,fscal,fix0);
1258             fiy0             = _mm_macc_ps(dy01,fscal,fiy0);
1259             fiz0             = _mm_macc_ps(dz01,fscal,fiz0);
1260
1261             fjx1             = _mm_macc_ps(dx01,fscal,fjx1);
1262             fjy1             = _mm_macc_ps(dy01,fscal,fjy1);
1263             fjz1             = _mm_macc_ps(dz01,fscal,fjz1);
1264
1265             /**************************
1266              * CALCULATE INTERACTIONS *
1267              **************************/
1268
1269             /* COULOMB ELECTROSTATICS */
1270             velec            = _mm_mul_ps(qq02,rinv02);
1271             felec            = _mm_mul_ps(velec,rinvsq02);
1272
1273             fscal            = felec;
1274
1275              /* Update vectorial force */
1276             fix0             = _mm_macc_ps(dx02,fscal,fix0);
1277             fiy0             = _mm_macc_ps(dy02,fscal,fiy0);
1278             fiz0             = _mm_macc_ps(dz02,fscal,fiz0);
1279
1280             fjx2             = _mm_macc_ps(dx02,fscal,fjx2);
1281             fjy2             = _mm_macc_ps(dy02,fscal,fjy2);
1282             fjz2             = _mm_macc_ps(dz02,fscal,fjz2);
1283
1284             /**************************
1285              * CALCULATE INTERACTIONS *
1286              **************************/
1287
1288             /* COULOMB ELECTROSTATICS */
1289             velec            = _mm_mul_ps(qq10,rinv10);
1290             felec            = _mm_mul_ps(velec,rinvsq10);
1291
1292             fscal            = felec;
1293
1294              /* Update vectorial force */
1295             fix1             = _mm_macc_ps(dx10,fscal,fix1);
1296             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
1297             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
1298
1299             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
1300             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
1301             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
1302
1303             /**************************
1304              * CALCULATE INTERACTIONS *
1305              **************************/
1306
1307             /* COULOMB ELECTROSTATICS */
1308             velec            = _mm_mul_ps(qq11,rinv11);
1309             felec            = _mm_mul_ps(velec,rinvsq11);
1310
1311             fscal            = felec;
1312
1313              /* Update vectorial force */
1314             fix1             = _mm_macc_ps(dx11,fscal,fix1);
1315             fiy1             = _mm_macc_ps(dy11,fscal,fiy1);
1316             fiz1             = _mm_macc_ps(dz11,fscal,fiz1);
1317
1318             fjx1             = _mm_macc_ps(dx11,fscal,fjx1);
1319             fjy1             = _mm_macc_ps(dy11,fscal,fjy1);
1320             fjz1             = _mm_macc_ps(dz11,fscal,fjz1);
1321
1322             /**************************
1323              * CALCULATE INTERACTIONS *
1324              **************************/
1325
1326             /* COULOMB ELECTROSTATICS */
1327             velec            = _mm_mul_ps(qq12,rinv12);
1328             felec            = _mm_mul_ps(velec,rinvsq12);
1329
1330             fscal            = felec;
1331
1332              /* Update vectorial force */
1333             fix1             = _mm_macc_ps(dx12,fscal,fix1);
1334             fiy1             = _mm_macc_ps(dy12,fscal,fiy1);
1335             fiz1             = _mm_macc_ps(dz12,fscal,fiz1);
1336
1337             fjx2             = _mm_macc_ps(dx12,fscal,fjx2);
1338             fjy2             = _mm_macc_ps(dy12,fscal,fjy2);
1339             fjz2             = _mm_macc_ps(dz12,fscal,fjz2);
1340
1341             /**************************
1342              * CALCULATE INTERACTIONS *
1343              **************************/
1344
1345             /* COULOMB ELECTROSTATICS */
1346             velec            = _mm_mul_ps(qq20,rinv20);
1347             felec            = _mm_mul_ps(velec,rinvsq20);
1348
1349             fscal            = felec;
1350
1351              /* Update vectorial force */
1352             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1353             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1354             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1355
1356             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1357             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1358             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1359
1360             /**************************
1361              * CALCULATE INTERACTIONS *
1362              **************************/
1363
1364             /* COULOMB ELECTROSTATICS */
1365             velec            = _mm_mul_ps(qq21,rinv21);
1366             felec            = _mm_mul_ps(velec,rinvsq21);
1367
1368             fscal            = felec;
1369
1370              /* Update vectorial force */
1371             fix2             = _mm_macc_ps(dx21,fscal,fix2);
1372             fiy2             = _mm_macc_ps(dy21,fscal,fiy2);
1373             fiz2             = _mm_macc_ps(dz21,fscal,fiz2);
1374
1375             fjx1             = _mm_macc_ps(dx21,fscal,fjx1);
1376             fjy1             = _mm_macc_ps(dy21,fscal,fjy1);
1377             fjz1             = _mm_macc_ps(dz21,fscal,fjz1);
1378
1379             /**************************
1380              * CALCULATE INTERACTIONS *
1381              **************************/
1382
1383             /* COULOMB ELECTROSTATICS */
1384             velec            = _mm_mul_ps(qq22,rinv22);
1385             felec            = _mm_mul_ps(velec,rinvsq22);
1386
1387             fscal            = felec;
1388
1389              /* Update vectorial force */
1390             fix2             = _mm_macc_ps(dx22,fscal,fix2);
1391             fiy2             = _mm_macc_ps(dy22,fscal,fiy2);
1392             fiz2             = _mm_macc_ps(dz22,fscal,fiz2);
1393
1394             fjx2             = _mm_macc_ps(dx22,fscal,fjx2);
1395             fjy2             = _mm_macc_ps(dy22,fscal,fjy2);
1396             fjz2             = _mm_macc_ps(dz22,fscal,fjz2);
1397
1398             fjptrA             = f+j_coord_offsetA;
1399             fjptrB             = f+j_coord_offsetB;
1400             fjptrC             = f+j_coord_offsetC;
1401             fjptrD             = f+j_coord_offsetD;
1402
1403             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1404                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1405
1406             /* Inner loop uses 297 flops */
1407         }
1408
1409         if(jidx<j_index_end)
1410         {
1411
1412             /* Get j neighbor index, and coordinate index */
1413             jnrlistA         = jjnr[jidx];
1414             jnrlistB         = jjnr[jidx+1];
1415             jnrlistC         = jjnr[jidx+2];
1416             jnrlistD         = jjnr[jidx+3];
1417             /* Sign of each element will be negative for non-real atoms.
1418              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1419              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1420              */
1421             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1422             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1423             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1424             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1425             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1426             j_coord_offsetA  = DIM*jnrA;
1427             j_coord_offsetB  = DIM*jnrB;
1428             j_coord_offsetC  = DIM*jnrC;
1429             j_coord_offsetD  = DIM*jnrD;
1430
1431             /* load j atom coordinates */
1432             gmx_mm_load_3rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1433                                               x+j_coord_offsetC,x+j_coord_offsetD,
1434                                               &jx0,&jy0,&jz0,&jx1,&jy1,&jz1,&jx2,&jy2,&jz2);
1435
1436             /* Calculate displacement vector */
1437             dx00             = _mm_sub_ps(ix0,jx0);
1438             dy00             = _mm_sub_ps(iy0,jy0);
1439             dz00             = _mm_sub_ps(iz0,jz0);
1440             dx01             = _mm_sub_ps(ix0,jx1);
1441             dy01             = _mm_sub_ps(iy0,jy1);
1442             dz01             = _mm_sub_ps(iz0,jz1);
1443             dx02             = _mm_sub_ps(ix0,jx2);
1444             dy02             = _mm_sub_ps(iy0,jy2);
1445             dz02             = _mm_sub_ps(iz0,jz2);
1446             dx10             = _mm_sub_ps(ix1,jx0);
1447             dy10             = _mm_sub_ps(iy1,jy0);
1448             dz10             = _mm_sub_ps(iz1,jz0);
1449             dx11             = _mm_sub_ps(ix1,jx1);
1450             dy11             = _mm_sub_ps(iy1,jy1);
1451             dz11             = _mm_sub_ps(iz1,jz1);
1452             dx12             = _mm_sub_ps(ix1,jx2);
1453             dy12             = _mm_sub_ps(iy1,jy2);
1454             dz12             = _mm_sub_ps(iz1,jz2);
1455             dx20             = _mm_sub_ps(ix2,jx0);
1456             dy20             = _mm_sub_ps(iy2,jy0);
1457             dz20             = _mm_sub_ps(iz2,jz0);
1458             dx21             = _mm_sub_ps(ix2,jx1);
1459             dy21             = _mm_sub_ps(iy2,jy1);
1460             dz21             = _mm_sub_ps(iz2,jz1);
1461             dx22             = _mm_sub_ps(ix2,jx2);
1462             dy22             = _mm_sub_ps(iy2,jy2);
1463             dz22             = _mm_sub_ps(iz2,jz2);
1464
1465             /* Calculate squared distance and things based on it */
1466             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1467             rsq01            = gmx_mm_calc_rsq_ps(dx01,dy01,dz01);
1468             rsq02            = gmx_mm_calc_rsq_ps(dx02,dy02,dz02);
1469             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1470             rsq11            = gmx_mm_calc_rsq_ps(dx11,dy11,dz11);
1471             rsq12            = gmx_mm_calc_rsq_ps(dx12,dy12,dz12);
1472             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1473             rsq21            = gmx_mm_calc_rsq_ps(dx21,dy21,dz21);
1474             rsq22            = gmx_mm_calc_rsq_ps(dx22,dy22,dz22);
1475
1476             rinv00           = avx128fma_invsqrt_f(rsq00);
1477             rinv01           = avx128fma_invsqrt_f(rsq01);
1478             rinv02           = avx128fma_invsqrt_f(rsq02);
1479             rinv10           = avx128fma_invsqrt_f(rsq10);
1480             rinv11           = avx128fma_invsqrt_f(rsq11);
1481             rinv12           = avx128fma_invsqrt_f(rsq12);
1482             rinv20           = avx128fma_invsqrt_f(rsq20);
1483             rinv21           = avx128fma_invsqrt_f(rsq21);
1484             rinv22           = avx128fma_invsqrt_f(rsq22);
1485
1486             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1487             rinvsq01         = _mm_mul_ps(rinv01,rinv01);
1488             rinvsq02         = _mm_mul_ps(rinv02,rinv02);
1489             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1490             rinvsq11         = _mm_mul_ps(rinv11,rinv11);
1491             rinvsq12         = _mm_mul_ps(rinv12,rinv12);
1492             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1493             rinvsq21         = _mm_mul_ps(rinv21,rinv21);
1494             rinvsq22         = _mm_mul_ps(rinv22,rinv22);
1495
1496             fjx0             = _mm_setzero_ps();
1497             fjy0             = _mm_setzero_ps();
1498             fjz0             = _mm_setzero_ps();
1499             fjx1             = _mm_setzero_ps();
1500             fjy1             = _mm_setzero_ps();
1501             fjz1             = _mm_setzero_ps();
1502             fjx2             = _mm_setzero_ps();
1503             fjy2             = _mm_setzero_ps();
1504             fjz2             = _mm_setzero_ps();
1505
1506             /**************************
1507              * CALCULATE INTERACTIONS *
1508              **************************/
1509
1510             r00              = _mm_mul_ps(rsq00,rinv00);
1511             r00              = _mm_andnot_ps(dummy_mask,r00);
1512
1513             /* Calculate table index by multiplying r with table scale and truncate to integer */
1514             rt               = _mm_mul_ps(r00,vftabscale);
1515             vfitab           = _mm_cvttps_epi32(rt);
1516 #ifdef __XOP__
1517             vfeps            = _mm_frcz_ps(rt);
1518 #else
1519             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
1520 #endif
1521             twovfeps         = _mm_add_ps(vfeps,vfeps);
1522             vfitab           = _mm_slli_epi32(vfitab,3);
1523
1524             /* COULOMB ELECTROSTATICS */
1525             velec            = _mm_mul_ps(qq00,rinv00);
1526             felec            = _mm_mul_ps(velec,rinvsq00);
1527
1528             /* CUBIC SPLINE TABLE DISPERSION */
1529             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1530             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1531             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1532             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1533             _MM_TRANSPOSE4_PS(Y,F,G,H);
1534             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1535             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1536             fvdw6            = _mm_mul_ps(c6_00,FF);
1537
1538             /* CUBIC SPLINE TABLE REPULSION */
1539             vfitab           = _mm_add_epi32(vfitab,ifour);
1540             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1541             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1542             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1543             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1544             _MM_TRANSPOSE4_PS(Y,F,G,H);
1545             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1546             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1547             fvdw12           = _mm_mul_ps(c12_00,FF);
1548             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
1549
1550             fscal            = _mm_add_ps(felec,fvdw);
1551
1552             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1553
1554              /* Update vectorial force */
1555             fix0             = _mm_macc_ps(dx00,fscal,fix0);
1556             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
1557             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
1558
1559             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
1560             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
1561             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
1562
1563             /**************************
1564              * CALCULATE INTERACTIONS *
1565              **************************/
1566
1567             /* COULOMB ELECTROSTATICS */
1568             velec            = _mm_mul_ps(qq01,rinv01);
1569             felec            = _mm_mul_ps(velec,rinvsq01);
1570
1571             fscal            = felec;
1572
1573             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1574
1575              /* Update vectorial force */
1576             fix0             = _mm_macc_ps(dx01,fscal,fix0);
1577             fiy0             = _mm_macc_ps(dy01,fscal,fiy0);
1578             fiz0             = _mm_macc_ps(dz01,fscal,fiz0);
1579
1580             fjx1             = _mm_macc_ps(dx01,fscal,fjx1);
1581             fjy1             = _mm_macc_ps(dy01,fscal,fjy1);
1582             fjz1             = _mm_macc_ps(dz01,fscal,fjz1);
1583
1584             /**************************
1585              * CALCULATE INTERACTIONS *
1586              **************************/
1587
1588             /* COULOMB ELECTROSTATICS */
1589             velec            = _mm_mul_ps(qq02,rinv02);
1590             felec            = _mm_mul_ps(velec,rinvsq02);
1591
1592             fscal            = felec;
1593
1594             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1595
1596              /* Update vectorial force */
1597             fix0             = _mm_macc_ps(dx02,fscal,fix0);
1598             fiy0             = _mm_macc_ps(dy02,fscal,fiy0);
1599             fiz0             = _mm_macc_ps(dz02,fscal,fiz0);
1600
1601             fjx2             = _mm_macc_ps(dx02,fscal,fjx2);
1602             fjy2             = _mm_macc_ps(dy02,fscal,fjy2);
1603             fjz2             = _mm_macc_ps(dz02,fscal,fjz2);
1604
1605             /**************************
1606              * CALCULATE INTERACTIONS *
1607              **************************/
1608
1609             /* COULOMB ELECTROSTATICS */
1610             velec            = _mm_mul_ps(qq10,rinv10);
1611             felec            = _mm_mul_ps(velec,rinvsq10);
1612
1613             fscal            = felec;
1614
1615             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1616
1617              /* Update vectorial force */
1618             fix1             = _mm_macc_ps(dx10,fscal,fix1);
1619             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
1620             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
1621
1622             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
1623             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
1624             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
1625
1626             /**************************
1627              * CALCULATE INTERACTIONS *
1628              **************************/
1629
1630             /* COULOMB ELECTROSTATICS */
1631             velec            = _mm_mul_ps(qq11,rinv11);
1632             felec            = _mm_mul_ps(velec,rinvsq11);
1633
1634             fscal            = felec;
1635
1636             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1637
1638              /* Update vectorial force */
1639             fix1             = _mm_macc_ps(dx11,fscal,fix1);
1640             fiy1             = _mm_macc_ps(dy11,fscal,fiy1);
1641             fiz1             = _mm_macc_ps(dz11,fscal,fiz1);
1642
1643             fjx1             = _mm_macc_ps(dx11,fscal,fjx1);
1644             fjy1             = _mm_macc_ps(dy11,fscal,fjy1);
1645             fjz1             = _mm_macc_ps(dz11,fscal,fjz1);
1646
1647             /**************************
1648              * CALCULATE INTERACTIONS *
1649              **************************/
1650
1651             /* COULOMB ELECTROSTATICS */
1652             velec            = _mm_mul_ps(qq12,rinv12);
1653             felec            = _mm_mul_ps(velec,rinvsq12);
1654
1655             fscal            = felec;
1656
1657             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1658
1659              /* Update vectorial force */
1660             fix1             = _mm_macc_ps(dx12,fscal,fix1);
1661             fiy1             = _mm_macc_ps(dy12,fscal,fiy1);
1662             fiz1             = _mm_macc_ps(dz12,fscal,fiz1);
1663
1664             fjx2             = _mm_macc_ps(dx12,fscal,fjx2);
1665             fjy2             = _mm_macc_ps(dy12,fscal,fjy2);
1666             fjz2             = _mm_macc_ps(dz12,fscal,fjz2);
1667
1668             /**************************
1669              * CALCULATE INTERACTIONS *
1670              **************************/
1671
1672             /* COULOMB ELECTROSTATICS */
1673             velec            = _mm_mul_ps(qq20,rinv20);
1674             felec            = _mm_mul_ps(velec,rinvsq20);
1675
1676             fscal            = felec;
1677
1678             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1679
1680              /* Update vectorial force */
1681             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1682             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1683             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1684
1685             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1686             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1687             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1688
1689             /**************************
1690              * CALCULATE INTERACTIONS *
1691              **************************/
1692
1693             /* COULOMB ELECTROSTATICS */
1694             velec            = _mm_mul_ps(qq21,rinv21);
1695             felec            = _mm_mul_ps(velec,rinvsq21);
1696
1697             fscal            = felec;
1698
1699             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1700
1701              /* Update vectorial force */
1702             fix2             = _mm_macc_ps(dx21,fscal,fix2);
1703             fiy2             = _mm_macc_ps(dy21,fscal,fiy2);
1704             fiz2             = _mm_macc_ps(dz21,fscal,fiz2);
1705
1706             fjx1             = _mm_macc_ps(dx21,fscal,fjx1);
1707             fjy1             = _mm_macc_ps(dy21,fscal,fjy1);
1708             fjz1             = _mm_macc_ps(dz21,fscal,fjz1);
1709
1710             /**************************
1711              * CALCULATE INTERACTIONS *
1712              **************************/
1713
1714             /* COULOMB ELECTROSTATICS */
1715             velec            = _mm_mul_ps(qq22,rinv22);
1716             felec            = _mm_mul_ps(velec,rinvsq22);
1717
1718             fscal            = felec;
1719
1720             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1721
1722              /* Update vectorial force */
1723             fix2             = _mm_macc_ps(dx22,fscal,fix2);
1724             fiy2             = _mm_macc_ps(dy22,fscal,fiy2);
1725             fiz2             = _mm_macc_ps(dz22,fscal,fiz2);
1726
1727             fjx2             = _mm_macc_ps(dx22,fscal,fjx2);
1728             fjy2             = _mm_macc_ps(dy22,fscal,fjy2);
1729             fjz2             = _mm_macc_ps(dz22,fscal,fjz2);
1730
1731             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1732             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1733             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1734             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1735
1736             gmx_mm_decrement_3rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
1737                                                    fjx0,fjy0,fjz0,fjx1,fjy1,fjz1,fjx2,fjy2,fjz2);
1738
1739             /* Inner loop uses 298 flops */
1740         }
1741
1742         /* End of innermost loop */
1743
1744         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1745                                               f+i_coord_offset,fshift+i_shift_offset);
1746
1747         /* Increment number of inner iterations */
1748         inneriter                  += j_index_end - j_index_start;
1749
1750         /* Outer loop uses 18 flops */
1751     }
1752
1753     /* Increment number of outer iterations */
1754     outeriter        += nri;
1755
1756     /* Update outer/inner flops */
1757
1758     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3W3_F,outeriter*18 + inneriter*298);
1759 }