src/gromacs/gmxlib/nonbonded/nb_kernel_avx_128_fma_single/nb_kernel_ElecCoul_VdwCSTab_GeomW4P1_avx_128_fma_single.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014,2015,2017,2018, 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_GeomW4P1_VF_avx_128_fma_single
  51  * Electrostatics interaction: Coulomb
  52  * VdW interaction:            CubicSplineTable
  53  * Geometry:                   Water4-Particle
  54  * Calculate force/pot:        PotentialAndForce
  55  */
  56 void
  57 nb_kernel_ElecCoul_VdwCSTab_GeomW4P1_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              vdwioffset3;
  89     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  90     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  91     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  92     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  93     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  94     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  95     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
  96     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
  97     real             *charge;
  98     int              nvdwtype;
  99     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 100     int              *vdwtype;
 101     real             *vdwparam;
 102     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 103     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 104     __m128i          vfitab;
 105     __m128i          ifour       = _mm_set1_epi32(4);
 106     __m128           rt,vfeps,twovfeps,vftabscale,Y,F,G,H,Fp,VV,FF;
 107     real             *vftab;
 108     __m128           dummy_mask,cutoff_mask;
 109     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 110     __m128           one     = _mm_set1_ps(1.0);
 111     __m128           two     = _mm_set1_ps(2.0);
 112     x                = xx[0];
 113     f                = ff[0];
 114
 115     nri              = nlist->nri;
 116     iinr             = nlist->iinr;
 117     jindex           = nlist->jindex;
 118     jjnr             = nlist->jjnr;
 119     shiftidx         = nlist->shift;
 120     gid              = nlist->gid;
 121     shiftvec         = fr->shift_vec[0];
 122     fshift           = fr->fshift[0];
 123     facel            = _mm_set1_ps(fr->ic->epsfac);
 124     charge           = mdatoms->chargeA;
 125     nvdwtype         = fr->ntype;
 126     vdwparam         = fr->nbfp;
 127     vdwtype          = mdatoms->typeA;
 128
 129     vftab            = kernel_data->table_vdw->data;
 130     vftabscale       = _mm_set1_ps(kernel_data->table_vdw->scale);
 131
 132     /* Setup water-specific parameters */
 133     inr              = nlist->iinr[0];
 134     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 135     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 136     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 137     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 138
 139     /* Avoid stupid compiler warnings */
 140     jnrA = jnrB = jnrC = jnrD = 0;
 141     j_coord_offsetA = 0;
 142     j_coord_offsetB = 0;
 143     j_coord_offsetC = 0;
 144     j_coord_offsetD = 0;
 145
 146     outeriter        = 0;
 147     inneriter        = 0;
 148
 149     for(iidx=0;iidx<4*DIM;iidx++)
 150     {
 151         scratch[iidx] = 0.0;
 152     }
 153
 154     /* Start outer loop over neighborlists */
 155     for(iidx=0; iidx<nri; iidx++)
 156     {
 157         /* Load shift vector for this list */
 158         i_shift_offset   = DIM*shiftidx[iidx];
 159
 160         /* Load limits for loop over neighbors */
 161         j_index_start    = jindex[iidx];
 162         j_index_end      = jindex[iidx+1];
 163
 164         /* Get outer coordinate index */
 165         inr              = iinr[iidx];
 166         i_coord_offset   = DIM*inr;
 167
 168         /* Load i particle coords and add shift vector */
 169         gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 170                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 171
 172         fix0             = _mm_setzero_ps();
 173         fiy0             = _mm_setzero_ps();
 174         fiz0             = _mm_setzero_ps();
 175         fix1             = _mm_setzero_ps();
 176         fiy1             = _mm_setzero_ps();
 177         fiz1             = _mm_setzero_ps();
 178         fix2             = _mm_setzero_ps();
 179         fiy2             = _mm_setzero_ps();
 180         fiz2             = _mm_setzero_ps();
 181         fix3             = _mm_setzero_ps();
 182         fiy3             = _mm_setzero_ps();
 183         fiz3             = _mm_setzero_ps();
 184
 185         /* Reset potential sums */
 186         velecsum         = _mm_setzero_ps();
 187         vvdwsum          = _mm_setzero_ps();
 188
 189         /* Start inner kernel loop */
 190         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 191         {
 192
 193             /* Get j neighbor index, and coordinate index */
 194             jnrA             = jjnr[jidx];
 195             jnrB             = jjnr[jidx+1];
 196             jnrC             = jjnr[jidx+2];
 197             jnrD             = jjnr[jidx+3];
 198             j_coord_offsetA  = DIM*jnrA;
 199             j_coord_offsetB  = DIM*jnrB;
 200             j_coord_offsetC  = DIM*jnrC;
 201             j_coord_offsetD  = DIM*jnrD;
 202
 203             /* load j atom coordinates */
 204             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 205                                               x+j_coord_offsetC,x+j_coord_offsetD,
 206                                               &jx0,&jy0,&jz0);
 207
 208             /* Calculate displacement vector */
 209             dx00             = _mm_sub_ps(ix0,jx0);
 210             dy00             = _mm_sub_ps(iy0,jy0);
 211             dz00             = _mm_sub_ps(iz0,jz0);
 212             dx10             = _mm_sub_ps(ix1,jx0);
 213             dy10             = _mm_sub_ps(iy1,jy0);
 214             dz10             = _mm_sub_ps(iz1,jz0);
 215             dx20             = _mm_sub_ps(ix2,jx0);
 216             dy20             = _mm_sub_ps(iy2,jy0);
 217             dz20             = _mm_sub_ps(iz2,jz0);
 218             dx30             = _mm_sub_ps(ix3,jx0);
 219             dy30             = _mm_sub_ps(iy3,jy0);
 220             dz30             = _mm_sub_ps(iz3,jz0);
 221
 222             /* Calculate squared distance and things based on it */
 223             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 224             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 225             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 226             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 227
 228             rinv00           = avx128fma_invsqrt_f(rsq00);
 229             rinv10           = avx128fma_invsqrt_f(rsq10);
 230             rinv20           = avx128fma_invsqrt_f(rsq20);
 231             rinv30           = avx128fma_invsqrt_f(rsq30);
 232
 233             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 234             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 235             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 236
 237             /* Load parameters for j particles */
 238             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 239                                                               charge+jnrC+0,charge+jnrD+0);
 240             vdwjidx0A        = 2*vdwtype[jnrA+0];
 241             vdwjidx0B        = 2*vdwtype[jnrB+0];
 242             vdwjidx0C        = 2*vdwtype[jnrC+0];
 243             vdwjidx0D        = 2*vdwtype[jnrD+0];
 244
 245             fjx0             = _mm_setzero_ps();
 246             fjy0             = _mm_setzero_ps();
 247             fjz0             = _mm_setzero_ps();
 248
 249             /**************************
 250              * CALCULATE INTERACTIONS *
 251              **************************/
 252
 253             r00              = _mm_mul_ps(rsq00,rinv00);
 254
 255             /* Compute parameters for interactions between i and j atoms */
 256             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 257                                          vdwparam+vdwioffset0+vdwjidx0B,
 258                                          vdwparam+vdwioffset0+vdwjidx0C,
 259                                          vdwparam+vdwioffset0+vdwjidx0D,
 260                                          &c6_00,&c12_00);
 261
 262             /* Calculate table index by multiplying r with table scale and truncate to integer */
 263             rt               = _mm_mul_ps(r00,vftabscale);
 264             vfitab           = _mm_cvttps_epi32(rt);
 265 #ifdef __XOP__
 266             vfeps            = _mm_frcz_ps(rt);
 267 #else
 268             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
 269 #endif
 270             twovfeps         = _mm_add_ps(vfeps,vfeps);
 271             vfitab           = _mm_slli_epi32(vfitab,3);
 272
 273             /* CUBIC SPLINE TABLE DISPERSION */
 274             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 275             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 276             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 277             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 278             _MM_TRANSPOSE4_PS(Y,F,G,H);
 279             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 280             VV               = _mm_macc_ps(vfeps,Fp,Y);
 281             vvdw6            = _mm_mul_ps(c6_00,VV);
 282             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 283             fvdw6            = _mm_mul_ps(c6_00,FF);
 284
 285             /* CUBIC SPLINE TABLE REPULSION */
 286             vfitab           = _mm_add_epi32(vfitab,ifour);
 287             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 288             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 289             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 290             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 291             _MM_TRANSPOSE4_PS(Y,F,G,H);
 292             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 293             VV               = _mm_macc_ps(vfeps,Fp,Y);
 294             vvdw12           = _mm_mul_ps(c12_00,VV);
 295             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 296             fvdw12           = _mm_mul_ps(c12_00,FF);
 297             vvdw             = _mm_add_ps(vvdw12,vvdw6);
 298             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
 299
 300             /* Update potential sum for this i atom from the interaction with this j atom. */
 301             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 302
 303             fscal            = fvdw;
 304
 305              /* Update vectorial force */
 306             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 307             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 308             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 309
 310             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 311             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 312             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 313
 314             /**************************
 315              * CALCULATE INTERACTIONS *
 316              **************************/
 317
 318             /* Compute parameters for interactions between i and j atoms */
 319             qq10             = _mm_mul_ps(iq1,jq0);
 320
 321             /* COULOMB ELECTROSTATICS */
 322             velec            = _mm_mul_ps(qq10,rinv10);
 323             felec            = _mm_mul_ps(velec,rinvsq10);
 324
 325             /* Update potential sum for this i atom from the interaction with this j atom. */
 326             velecsum         = _mm_add_ps(velecsum,velec);
 327
 328             fscal            = felec;
 329
 330              /* Update vectorial force */
 331             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 332             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 333             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 334
 335             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 336             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 337             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 338
 339             /**************************
 340              * CALCULATE INTERACTIONS *
 341              **************************/
 342
 343             /* Compute parameters for interactions between i and j atoms */
 344             qq20             = _mm_mul_ps(iq2,jq0);
 345
 346             /* COULOMB ELECTROSTATICS */
 347             velec            = _mm_mul_ps(qq20,rinv20);
 348             felec            = _mm_mul_ps(velec,rinvsq20);
 349
 350             /* Update potential sum for this i atom from the interaction with this j atom. */
 351             velecsum         = _mm_add_ps(velecsum,velec);
 352
 353             fscal            = felec;
 354
 355              /* Update vectorial force */
 356             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 357             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 358             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 359
 360             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 361             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 362             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 363
 364             /**************************
 365              * CALCULATE INTERACTIONS *
 366              **************************/
 367
 368             /* Compute parameters for interactions between i and j atoms */
 369             qq30             = _mm_mul_ps(iq3,jq0);
 370
 371             /* COULOMB ELECTROSTATICS */
 372             velec            = _mm_mul_ps(qq30,rinv30);
 373             felec            = _mm_mul_ps(velec,rinvsq30);
 374
 375             /* Update potential sum for this i atom from the interaction with this j atom. */
 376             velecsum         = _mm_add_ps(velecsum,velec);
 377
 378             fscal            = felec;
 379
 380              /* Update vectorial force */
 381             fix3             = _mm_macc_ps(dx30,fscal,fix3);
 382             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
 383             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
 384
 385             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
 386             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
 387             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
 388
 389             fjptrA             = f+j_coord_offsetA;
 390             fjptrB             = f+j_coord_offsetB;
 391             fjptrC             = f+j_coord_offsetC;
 392             fjptrD             = f+j_coord_offsetD;
 393
 394             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 395
 396             /* Inner loop uses 152 flops */
 397         }
 398
 399         if(jidx<j_index_end)
 400         {
 401
 402             /* Get j neighbor index, and coordinate index */
 403             jnrlistA         = jjnr[jidx];
 404             jnrlistB         = jjnr[jidx+1];
 405             jnrlistC         = jjnr[jidx+2];
 406             jnrlistD         = jjnr[jidx+3];
 407             /* Sign of each element will be negative for non-real atoms.
 408              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 409              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 410              */
 411             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 412             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 413             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 414             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 415             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 416             j_coord_offsetA  = DIM*jnrA;
 417             j_coord_offsetB  = DIM*jnrB;
 418             j_coord_offsetC  = DIM*jnrC;
 419             j_coord_offsetD  = DIM*jnrD;
 420
 421             /* load j atom coordinates */
 422             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 423                                               x+j_coord_offsetC,x+j_coord_offsetD,
 424                                               &jx0,&jy0,&jz0);
 425
 426             /* Calculate displacement vector */
 427             dx00             = _mm_sub_ps(ix0,jx0);
 428             dy00             = _mm_sub_ps(iy0,jy0);
 429             dz00             = _mm_sub_ps(iz0,jz0);
 430             dx10             = _mm_sub_ps(ix1,jx0);
 431             dy10             = _mm_sub_ps(iy1,jy0);
 432             dz10             = _mm_sub_ps(iz1,jz0);
 433             dx20             = _mm_sub_ps(ix2,jx0);
 434             dy20             = _mm_sub_ps(iy2,jy0);
 435             dz20             = _mm_sub_ps(iz2,jz0);
 436             dx30             = _mm_sub_ps(ix3,jx0);
 437             dy30             = _mm_sub_ps(iy3,jy0);
 438             dz30             = _mm_sub_ps(iz3,jz0);
 439
 440             /* Calculate squared distance and things based on it */
 441             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 442             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 443             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 444             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 445
 446             rinv00           = avx128fma_invsqrt_f(rsq00);
 447             rinv10           = avx128fma_invsqrt_f(rsq10);
 448             rinv20           = avx128fma_invsqrt_f(rsq20);
 449             rinv30           = avx128fma_invsqrt_f(rsq30);
 450
 451             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 452             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 453             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 454
 455             /* Load parameters for j particles */
 456             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 457                                                               charge+jnrC+0,charge+jnrD+0);
 458             vdwjidx0A        = 2*vdwtype[jnrA+0];
 459             vdwjidx0B        = 2*vdwtype[jnrB+0];
 460             vdwjidx0C        = 2*vdwtype[jnrC+0];
 461             vdwjidx0D        = 2*vdwtype[jnrD+0];
 462
 463             fjx0             = _mm_setzero_ps();
 464             fjy0             = _mm_setzero_ps();
 465             fjz0             = _mm_setzero_ps();
 466
 467             /**************************
 468              * CALCULATE INTERACTIONS *
 469              **************************/
 470
 471             r00              = _mm_mul_ps(rsq00,rinv00);
 472             r00              = _mm_andnot_ps(dummy_mask,r00);
 473
 474             /* Compute parameters for interactions between i and j atoms */
 475             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 476                                          vdwparam+vdwioffset0+vdwjidx0B,
 477                                          vdwparam+vdwioffset0+vdwjidx0C,
 478                                          vdwparam+vdwioffset0+vdwjidx0D,
 479                                          &c6_00,&c12_00);
 480
 481             /* Calculate table index by multiplying r with table scale and truncate to integer */
 482             rt               = _mm_mul_ps(r00,vftabscale);
 483             vfitab           = _mm_cvttps_epi32(rt);
 484 #ifdef __XOP__
 485             vfeps            = _mm_frcz_ps(rt);
 486 #else
 487             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
 488 #endif
 489             twovfeps         = _mm_add_ps(vfeps,vfeps);
 490             vfitab           = _mm_slli_epi32(vfitab,3);
 491
 492             /* CUBIC SPLINE TABLE DISPERSION */
 493             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 494             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 495             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 496             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 497             _MM_TRANSPOSE4_PS(Y,F,G,H);
 498             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 499             VV               = _mm_macc_ps(vfeps,Fp,Y);
 500             vvdw6            = _mm_mul_ps(c6_00,VV);
 501             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 502             fvdw6            = _mm_mul_ps(c6_00,FF);
 503
 504             /* CUBIC SPLINE TABLE REPULSION */
 505             vfitab           = _mm_add_epi32(vfitab,ifour);
 506             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 507             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 508             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 509             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 510             _MM_TRANSPOSE4_PS(Y,F,G,H);
 511             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 512             VV               = _mm_macc_ps(vfeps,Fp,Y);
 513             vvdw12           = _mm_mul_ps(c12_00,VV);
 514             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 515             fvdw12           = _mm_mul_ps(c12_00,FF);
 516             vvdw             = _mm_add_ps(vvdw12,vvdw6);
 517             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
 518
 519             /* Update potential sum for this i atom from the interaction with this j atom. */
 520             vvdw             = _mm_andnot_ps(dummy_mask,vvdw);
 521             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 522
 523             fscal            = fvdw;
 524
 525             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 526
 527              /* Update vectorial force */
 528             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 529             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 530             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 531
 532             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 533             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 534             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 535
 536             /**************************
 537              * CALCULATE INTERACTIONS *
 538              **************************/
 539
 540             /* Compute parameters for interactions between i and j atoms */
 541             qq10             = _mm_mul_ps(iq1,jq0);
 542
 543             /* COULOMB ELECTROSTATICS */
 544             velec            = _mm_mul_ps(qq10,rinv10);
 545             felec            = _mm_mul_ps(velec,rinvsq10);
 546
 547             /* Update potential sum for this i atom from the interaction with this j atom. */
 548             velec            = _mm_andnot_ps(dummy_mask,velec);
 549             velecsum         = _mm_add_ps(velecsum,velec);
 550
 551             fscal            = felec;
 552
 553             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 554
 555              /* Update vectorial force */
 556             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 557             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 558             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 559
 560             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 561             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 562             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 563
 564             /**************************
 565              * CALCULATE INTERACTIONS *
 566              **************************/
 567
 568             /* Compute parameters for interactions between i and j atoms */
 569             qq20             = _mm_mul_ps(iq2,jq0);
 570
 571             /* COULOMB ELECTROSTATICS */
 572             velec            = _mm_mul_ps(qq20,rinv20);
 573             felec            = _mm_mul_ps(velec,rinvsq20);
 574
 575             /* Update potential sum for this i atom from the interaction with this j atom. */
 576             velec            = _mm_andnot_ps(dummy_mask,velec);
 577             velecsum         = _mm_add_ps(velecsum,velec);
 578
 579             fscal            = felec;
 580
 581             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 582
 583              /* Update vectorial force */
 584             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 585             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 586             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 587
 588             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 589             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 590             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 591
 592             /**************************
 593              * CALCULATE INTERACTIONS *
 594              **************************/
 595
 596             /* Compute parameters for interactions between i and j atoms */
 597             qq30             = _mm_mul_ps(iq3,jq0);
 598
 599             /* COULOMB ELECTROSTATICS */
 600             velec            = _mm_mul_ps(qq30,rinv30);
 601             felec            = _mm_mul_ps(velec,rinvsq30);
 602
 603             /* Update potential sum for this i atom from the interaction with this j atom. */
 604             velec            = _mm_andnot_ps(dummy_mask,velec);
 605             velecsum         = _mm_add_ps(velecsum,velec);
 606
 607             fscal            = felec;
 608
 609             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 610
 611              /* Update vectorial force */
 612             fix3             = _mm_macc_ps(dx30,fscal,fix3);
 613             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
 614             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
 615
 616             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
 617             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
 618             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
 619
 620             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 621             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 622             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 623             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 624
 625             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 626
 627             /* Inner loop uses 153 flops */
 628         }
 629
 630         /* End of innermost loop */
 631
 632         gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 633                                               f+i_coord_offset,fshift+i_shift_offset);
 634
 635         ggid                        = gid[iidx];
 636         /* Update potential energies */
 637         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 638         gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 639
 640         /* Increment number of inner iterations */
 641         inneriter                  += j_index_end - j_index_start;
 642
 643         /* Outer loop uses 26 flops */
 644     }
 645
 646     /* Increment number of outer iterations */
 647     outeriter        += nri;
 648
 649     /* Update outer/inner flops */
 650
 651     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_VF,outeriter*26 + inneriter*153);
 652 }
 653 /*
 654  * Gromacs nonbonded kernel:   nb_kernel_ElecCoul_VdwCSTab_GeomW4P1_F_avx_128_fma_single
 655  * Electrostatics interaction: Coulomb
 656  * VdW interaction:            CubicSplineTable
 657  * Geometry:                   Water4-Particle
 658  * Calculate force/pot:        Force
 659  */
 660 void
 661 nb_kernel_ElecCoul_VdwCSTab_GeomW4P1_F_avx_128_fma_single
 662                     (t_nblist                    * gmx_restrict       nlist,
 663                      rvec                        * gmx_restrict          xx,
 664                      rvec                        * gmx_restrict          ff,
 665                      struct t_forcerec           * gmx_restrict          fr,
 666                      t_mdatoms                   * gmx_restrict     mdatoms,
 667                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 668                      t_nrnb                      * gmx_restrict        nrnb)
 669 {
 670     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 671      * just 0 for non-waters.
 672      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
 673      * jnr indices corresponding to data put in the four positions in the SIMD register.
 674      */
 675     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 676     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 677     int              jnrA,jnrB,jnrC,jnrD;
 678     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 679     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 680     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 681     real             rcutoff_scalar;
 682     real             *shiftvec,*fshift,*x,*f;
 683     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 684     real             scratch[4*DIM];
 685     __m128           fscal,rcutoff,rcutoff2,jidxall;
 686     int              vdwioffset0;
 687     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 688     int              vdwioffset1;
 689     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 690     int              vdwioffset2;
 691     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 692     int              vdwioffset3;
 693     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 694     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 695     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 696     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 697     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 698     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 699     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 700     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 701     real             *charge;
 702     int              nvdwtype;
 703     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 704     int              *vdwtype;
 705     real             *vdwparam;
 706     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 707     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 708     __m128i          vfitab;
 709     __m128i          ifour       = _mm_set1_epi32(4);
 710     __m128           rt,vfeps,twovfeps,vftabscale,Y,F,G,H,Fp,VV,FF;
 711     real             *vftab;
 712     __m128           dummy_mask,cutoff_mask;
 713     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 714     __m128           one     = _mm_set1_ps(1.0);
 715     __m128           two     = _mm_set1_ps(2.0);
 716     x                = xx[0];
 717     f                = ff[0];
 718
 719     nri              = nlist->nri;
 720     iinr             = nlist->iinr;
 721     jindex           = nlist->jindex;
 722     jjnr             = nlist->jjnr;
 723     shiftidx         = nlist->shift;
 724     gid              = nlist->gid;
 725     shiftvec         = fr->shift_vec[0];
 726     fshift           = fr->fshift[0];
 727     facel            = _mm_set1_ps(fr->ic->epsfac);
 728     charge           = mdatoms->chargeA;
 729     nvdwtype         = fr->ntype;
 730     vdwparam         = fr->nbfp;
 731     vdwtype          = mdatoms->typeA;
 732
 733     vftab            = kernel_data->table_vdw->data;
 734     vftabscale       = _mm_set1_ps(kernel_data->table_vdw->scale);
 735
 736     /* Setup water-specific parameters */
 737     inr              = nlist->iinr[0];
 738     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 739     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 740     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 741     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 742
 743     /* Avoid stupid compiler warnings */
 744     jnrA = jnrB = jnrC = jnrD = 0;
 745     j_coord_offsetA = 0;
 746     j_coord_offsetB = 0;
 747     j_coord_offsetC = 0;
 748     j_coord_offsetD = 0;
 749
 750     outeriter        = 0;
 751     inneriter        = 0;
 752
 753     for(iidx=0;iidx<4*DIM;iidx++)
 754     {
 755         scratch[iidx] = 0.0;
 756     }
 757
 758     /* Start outer loop over neighborlists */
 759     for(iidx=0; iidx<nri; iidx++)
 760     {
 761         /* Load shift vector for this list */
 762         i_shift_offset   = DIM*shiftidx[iidx];
 763
 764         /* Load limits for loop over neighbors */
 765         j_index_start    = jindex[iidx];
 766         j_index_end      = jindex[iidx+1];
 767
 768         /* Get outer coordinate index */
 769         inr              = iinr[iidx];
 770         i_coord_offset   = DIM*inr;
 771
 772         /* Load i particle coords and add shift vector */
 773         gmx_mm_load_shift_and_4rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 774                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 775
 776         fix0             = _mm_setzero_ps();
 777         fiy0             = _mm_setzero_ps();
 778         fiz0             = _mm_setzero_ps();
 779         fix1             = _mm_setzero_ps();
 780         fiy1             = _mm_setzero_ps();
 781         fiz1             = _mm_setzero_ps();
 782         fix2             = _mm_setzero_ps();
 783         fiy2             = _mm_setzero_ps();
 784         fiz2             = _mm_setzero_ps();
 785         fix3             = _mm_setzero_ps();
 786         fiy3             = _mm_setzero_ps();
 787         fiz3             = _mm_setzero_ps();
 788
 789         /* Start inner kernel loop */
 790         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 791         {
 792
 793             /* Get j neighbor index, and coordinate index */
 794             jnrA             = jjnr[jidx];
 795             jnrB             = jjnr[jidx+1];
 796             jnrC             = jjnr[jidx+2];
 797             jnrD             = jjnr[jidx+3];
 798             j_coord_offsetA  = DIM*jnrA;
 799             j_coord_offsetB  = DIM*jnrB;
 800             j_coord_offsetC  = DIM*jnrC;
 801             j_coord_offsetD  = DIM*jnrD;
 802
 803             /* load j atom coordinates */
 804             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 805                                               x+j_coord_offsetC,x+j_coord_offsetD,
 806                                               &jx0,&jy0,&jz0);
 807
 808             /* Calculate displacement vector */
 809             dx00             = _mm_sub_ps(ix0,jx0);
 810             dy00             = _mm_sub_ps(iy0,jy0);
 811             dz00             = _mm_sub_ps(iz0,jz0);
 812             dx10             = _mm_sub_ps(ix1,jx0);
 813             dy10             = _mm_sub_ps(iy1,jy0);
 814             dz10             = _mm_sub_ps(iz1,jz0);
 815             dx20             = _mm_sub_ps(ix2,jx0);
 816             dy20             = _mm_sub_ps(iy2,jy0);
 817             dz20             = _mm_sub_ps(iz2,jz0);
 818             dx30             = _mm_sub_ps(ix3,jx0);
 819             dy30             = _mm_sub_ps(iy3,jy0);
 820             dz30             = _mm_sub_ps(iz3,jz0);
 821
 822             /* Calculate squared distance and things based on it */
 823             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 824             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 825             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 826             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 827
 828             rinv00           = avx128fma_invsqrt_f(rsq00);
 829             rinv10           = avx128fma_invsqrt_f(rsq10);
 830             rinv20           = avx128fma_invsqrt_f(rsq20);
 831             rinv30           = avx128fma_invsqrt_f(rsq30);
 832
 833             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 834             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 835             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 836
 837             /* Load parameters for j particles */
 838             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 839                                                               charge+jnrC+0,charge+jnrD+0);
 840             vdwjidx0A        = 2*vdwtype[jnrA+0];
 841             vdwjidx0B        = 2*vdwtype[jnrB+0];
 842             vdwjidx0C        = 2*vdwtype[jnrC+0];
 843             vdwjidx0D        = 2*vdwtype[jnrD+0];
 844
 845             fjx0             = _mm_setzero_ps();
 846             fjy0             = _mm_setzero_ps();
 847             fjz0             = _mm_setzero_ps();
 848
 849             /**************************
 850              * CALCULATE INTERACTIONS *
 851              **************************/
 852
 853             r00              = _mm_mul_ps(rsq00,rinv00);
 854
 855             /* Compute parameters for interactions between i and j atoms */
 856             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 857                                          vdwparam+vdwioffset0+vdwjidx0B,
 858                                          vdwparam+vdwioffset0+vdwjidx0C,
 859                                          vdwparam+vdwioffset0+vdwjidx0D,
 860                                          &c6_00,&c12_00);
 861
 862             /* Calculate table index by multiplying r with table scale and truncate to integer */
 863             rt               = _mm_mul_ps(r00,vftabscale);
 864             vfitab           = _mm_cvttps_epi32(rt);
 865 #ifdef __XOP__
 866             vfeps            = _mm_frcz_ps(rt);
 867 #else
 868             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
 869 #endif
 870             twovfeps         = _mm_add_ps(vfeps,vfeps);
 871             vfitab           = _mm_slli_epi32(vfitab,3);
 872
 873             /* CUBIC SPLINE TABLE DISPERSION */
 874             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 875             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 876             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 877             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 878             _MM_TRANSPOSE4_PS(Y,F,G,H);
 879             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 880             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 881             fvdw6            = _mm_mul_ps(c6_00,FF);
 882
 883             /* CUBIC SPLINE TABLE REPULSION */
 884             vfitab           = _mm_add_epi32(vfitab,ifour);
 885             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
 886             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
 887             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
 888             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
 889             _MM_TRANSPOSE4_PS(Y,F,G,H);
 890             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
 891             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
 892             fvdw12           = _mm_mul_ps(c12_00,FF);
 893             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
 894
 895             fscal            = fvdw;
 896
 897              /* Update vectorial force */
 898             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 899             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 900             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 901
 902             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 903             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 904             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 905
 906             /**************************
 907              * CALCULATE INTERACTIONS *
 908              **************************/
 909
 910             /* Compute parameters for interactions between i and j atoms */
 911             qq10             = _mm_mul_ps(iq1,jq0);
 912
 913             /* COULOMB ELECTROSTATICS */
 914             velec            = _mm_mul_ps(qq10,rinv10);
 915             felec            = _mm_mul_ps(velec,rinvsq10);
 916
 917             fscal            = felec;
 918
 919              /* Update vectorial force */
 920             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 921             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 922             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 923
 924             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 925             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 926             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 927
 928             /**************************
 929              * CALCULATE INTERACTIONS *
 930              **************************/
 931
 932             /* Compute parameters for interactions between i and j atoms */
 933             qq20             = _mm_mul_ps(iq2,jq0);
 934
 935             /* COULOMB ELECTROSTATICS */
 936             velec            = _mm_mul_ps(qq20,rinv20);
 937             felec            = _mm_mul_ps(velec,rinvsq20);
 938
 939             fscal            = felec;
 940
 941              /* Update vectorial force */
 942             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 943             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 944             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 945
 946             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 947             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 948             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 949
 950             /**************************
 951              * CALCULATE INTERACTIONS *
 952              **************************/
 953
 954             /* Compute parameters for interactions between i and j atoms */
 955             qq30             = _mm_mul_ps(iq3,jq0);
 956
 957             /* COULOMB ELECTROSTATICS */
 958             velec            = _mm_mul_ps(qq30,rinv30);
 959             felec            = _mm_mul_ps(velec,rinvsq30);
 960
 961             fscal            = felec;
 962
 963              /* Update vectorial force */
 964             fix3             = _mm_macc_ps(dx30,fscal,fix3);
 965             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
 966             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
 967
 968             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
 969             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
 970             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
 971
 972             fjptrA             = f+j_coord_offsetA;
 973             fjptrB             = f+j_coord_offsetB;
 974             fjptrC             = f+j_coord_offsetC;
 975             fjptrD             = f+j_coord_offsetD;
 976
 977             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 978
 979             /* Inner loop uses 141 flops */
 980         }
 981
 982         if(jidx<j_index_end)
 983         {
 984
 985             /* Get j neighbor index, and coordinate index */
 986             jnrlistA         = jjnr[jidx];
 987             jnrlistB         = jjnr[jidx+1];
 988             jnrlistC         = jjnr[jidx+2];
 989             jnrlistD         = jjnr[jidx+3];
 990             /* Sign of each element will be negative for non-real atoms.
 991              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 992              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 993              */
 994             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 995             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 996             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 997             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 998             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 999             j_coord_offsetA  = DIM*jnrA;
1000             j_coord_offsetB  = DIM*jnrB;
1001             j_coord_offsetC  = DIM*jnrC;
1002             j_coord_offsetD  = DIM*jnrD;
1003
1004             /* load j atom coordinates */
1005             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1006                                               x+j_coord_offsetC,x+j_coord_offsetD,
1007                                               &jx0,&jy0,&jz0);
1008
1009             /* Calculate displacement vector */
1010             dx00             = _mm_sub_ps(ix0,jx0);
1011             dy00             = _mm_sub_ps(iy0,jy0);
1012             dz00             = _mm_sub_ps(iz0,jz0);
1013             dx10             = _mm_sub_ps(ix1,jx0);
1014             dy10             = _mm_sub_ps(iy1,jy0);
1015             dz10             = _mm_sub_ps(iz1,jz0);
1016             dx20             = _mm_sub_ps(ix2,jx0);
1017             dy20             = _mm_sub_ps(iy2,jy0);
1018             dz20             = _mm_sub_ps(iz2,jz0);
1019             dx30             = _mm_sub_ps(ix3,jx0);
1020             dy30             = _mm_sub_ps(iy3,jy0);
1021             dz30             = _mm_sub_ps(iz3,jz0);
1022
1023             /* Calculate squared distance and things based on it */
1024             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1025             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1026             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1027             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1028
1029             rinv00           = avx128fma_invsqrt_f(rsq00);
1030             rinv10           = avx128fma_invsqrt_f(rsq10);
1031             rinv20           = avx128fma_invsqrt_f(rsq20);
1032             rinv30           = avx128fma_invsqrt_f(rsq30);
1033
1034             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1035             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1036             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
1037
1038             /* Load parameters for j particles */
1039             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1040                                                               charge+jnrC+0,charge+jnrD+0);
1041             vdwjidx0A        = 2*vdwtype[jnrA+0];
1042             vdwjidx0B        = 2*vdwtype[jnrB+0];
1043             vdwjidx0C        = 2*vdwtype[jnrC+0];
1044             vdwjidx0D        = 2*vdwtype[jnrD+0];
1045
1046             fjx0             = _mm_setzero_ps();
1047             fjy0             = _mm_setzero_ps();
1048             fjz0             = _mm_setzero_ps();
1049
1050             /**************************
1051              * CALCULATE INTERACTIONS *
1052              **************************/
1053
1054             r00              = _mm_mul_ps(rsq00,rinv00);
1055             r00              = _mm_andnot_ps(dummy_mask,r00);
1056
1057             /* Compute parameters for interactions between i and j atoms */
1058             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
1059                                          vdwparam+vdwioffset0+vdwjidx0B,
1060                                          vdwparam+vdwioffset0+vdwjidx0C,
1061                                          vdwparam+vdwioffset0+vdwjidx0D,
1062                                          &c6_00,&c12_00);
1063
1064             /* Calculate table index by multiplying r with table scale and truncate to integer */
1065             rt               = _mm_mul_ps(r00,vftabscale);
1066             vfitab           = _mm_cvttps_epi32(rt);
1067 #ifdef __XOP__
1068             vfeps            = _mm_frcz_ps(rt);
1069 #else
1070             vfeps            = _mm_sub_ps(rt,_mm_round_ps(rt, _MM_FROUND_FLOOR));
1071 #endif
1072             twovfeps         = _mm_add_ps(vfeps,vfeps);
1073             vfitab           = _mm_slli_epi32(vfitab,3);
1074
1075             /* CUBIC SPLINE TABLE DISPERSION */
1076             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1077             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1078             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1079             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1080             _MM_TRANSPOSE4_PS(Y,F,G,H);
1081             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1082             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1083             fvdw6            = _mm_mul_ps(c6_00,FF);
1084
1085             /* CUBIC SPLINE TABLE REPULSION */
1086             vfitab           = _mm_add_epi32(vfitab,ifour);
1087             Y                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,0) );
1088             F                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,1) );
1089             G                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,2) );
1090             H                = _mm_load_ps( vftab + _mm_extract_epi32(vfitab,3) );
1091             _MM_TRANSPOSE4_PS(Y,F,G,H);
1092             Fp               = _mm_macc_ps(vfeps,_mm_macc_ps(H,vfeps,G),F);
1093             FF               = _mm_macc_ps(vfeps,_mm_macc_ps(twovfeps,H,G),Fp);
1094             fvdw12           = _mm_mul_ps(c12_00,FF);
1095             fvdw             = _mm_xor_ps(signbit,_mm_mul_ps(_mm_add_ps(fvdw6,fvdw12),_mm_mul_ps(vftabscale,rinv00)));
1096
1097             fscal            = fvdw;
1098
1099             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1100
1101              /* Update vectorial force */
1102             fix0             = _mm_macc_ps(dx00,fscal,fix0);
1103             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
1104             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
1105
1106             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
1107             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
1108             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
1109
1110             /**************************
1111              * CALCULATE INTERACTIONS *
1112              **************************/
1113
1114             /* Compute parameters for interactions between i and j atoms */
1115             qq10             = _mm_mul_ps(iq1,jq0);
1116
1117             /* COULOMB ELECTROSTATICS */
1118             velec            = _mm_mul_ps(qq10,rinv10);
1119             felec            = _mm_mul_ps(velec,rinvsq10);
1120
1121             fscal            = felec;
1122
1123             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1124
1125              /* Update vectorial force */
1126             fix1             = _mm_macc_ps(dx10,fscal,fix1);
1127             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
1128             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
1129
1130             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
1131             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
1132             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
1133
1134             /**************************
1135              * CALCULATE INTERACTIONS *
1136              **************************/
1137
1138             /* Compute parameters for interactions between i and j atoms */
1139             qq20             = _mm_mul_ps(iq2,jq0);
1140
1141             /* COULOMB ELECTROSTATICS */
1142             velec            = _mm_mul_ps(qq20,rinv20);
1143             felec            = _mm_mul_ps(velec,rinvsq20);
1144
1145             fscal            = felec;
1146
1147             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1148
1149              /* Update vectorial force */
1150             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1151             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1152             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1153
1154             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1155             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1156             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1157
1158             /**************************
1159              * CALCULATE INTERACTIONS *
1160              **************************/
1161
1162             /* Compute parameters for interactions between i and j atoms */
1163             qq30             = _mm_mul_ps(iq3,jq0);
1164
1165             /* COULOMB ELECTROSTATICS */
1166             velec            = _mm_mul_ps(qq30,rinv30);
1167             felec            = _mm_mul_ps(velec,rinvsq30);
1168
1169             fscal            = felec;
1170
1171             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1172
1173              /* Update vectorial force */
1174             fix3             = _mm_macc_ps(dx30,fscal,fix3);
1175             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
1176             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
1177
1178             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
1179             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
1180             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
1181
1182             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1183             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1184             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1185             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1186
1187             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1188
1189             /* Inner loop uses 142 flops */
1190         }
1191
1192         /* End of innermost loop */
1193
1194         gmx_mm_update_iforce_4atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1195                                               f+i_coord_offset,fshift+i_shift_offset);
1196
1197         /* Increment number of inner iterations */
1198         inneriter                  += j_index_end - j_index_start;
1199
1200         /* Outer loop uses 24 flops */
1201     }
1202
1203     /* Increment number of outer iterations */
1204     outeriter        += nri;
1205
1206     /* Update outer/inner flops */
1207
1208     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4_F,outeriter*24 + inneriter*142);
1209 }