src/gromacs/gmxlib/nonbonded/nb_kernel_avx_128_fma_single/nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_avx_128_fma_single.c

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
   5  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   6  * and including many others, as listed in the AUTHORS file in the
   7  * top-level source directory and at http://www.gromacs.org.
   8  *
   9  * GROMACS is free software; you can redistribute it and/or
  10  * modify it under the terms of the GNU Lesser General Public License
  11  * as published by the Free Software Foundation; either version 2.1
  12  * of the License, or (at your option) any later version.
  13  *
  14  * GROMACS is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17  * Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public
  20  * License along with GROMACS; if not, see
  21  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  23  *
  24  * If you want to redistribute modifications to GROMACS, please
  25  * consider that scientific software is very special. Version
  26  * control is crucial - bugs must be traceable. We will be happy to
  27  * consider code for inclusion in the official distribution, but
  28  * derived work must not be called official GROMACS. Details are found
  29  * in the README & COPYING files - if they are missing, get the
  30  * official version at http://www.gromacs.org.
  31  *
  32  * To help us fund GROMACS development, we humbly ask that you cite
  33  * the research papers on the package. Check out http://www.gromacs.org.
  34  */
  35 /*
  36  * Note: this file was generated by the GROMACS avx_128_fma_single kernel generator.
  37  */
  38 #ifdef HAVE_CONFIG_H
  39 #include <config.h>
  40 #endif
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "types/simple.h"
  46 #include "vec.h"
  47 #include "nrnb.h"
  48
  49 #include "gromacs/simd/math_x86_avx_128_fma_single.h"
  50 #include "kernelutil_x86_avx_128_fma_single.h"
  51
  52 /*
  53  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_VF_avx_128_fma_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            LJEwald
  56  * Geometry:                   Particle-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_VF_avx_128_fma_single
  61                     (t_nblist                    * gmx_restrict       nlist,
  62                      rvec                        * gmx_restrict          xx,
  63                      rvec                        * gmx_restrict          ff,
  64                      t_forcerec                  * gmx_restrict          fr,
  65                      t_mdatoms                   * gmx_restrict     mdatoms,
  66                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  67                      t_nrnb                      * gmx_restrict        nrnb)
  68 {
  69     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  70      * just 0 for non-waters.
  71      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
  72      * jnr indices corresponding to data put in the four positions in the SIMD register.
  73      */
  74     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  75     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  76     int              jnrA,jnrB,jnrC,jnrD;
  77     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  78     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  79     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  80     real             rcutoff_scalar;
  81     real             *shiftvec,*fshift,*x,*f;
  82     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  83     real             scratch[4*DIM];
  84     __m128           fscal,rcutoff,rcutoff2,jidxall;
  85     int              vdwioffset0;
  86     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  87     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  88     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  89     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  90     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
  91     real             *charge;
  92     int              nvdwtype;
  93     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  94     int              *vdwtype;
  95     real             *vdwparam;
  96     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
  97     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
  98     __m128           c6grid_00;
  99     real             *vdwgridparam;
 100     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 101     __m128           one_half = _mm_set1_ps(0.5);
 102     __m128           minus_one = _mm_set1_ps(-1.0);
 103     __m128i          ewitab;
 104     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 105     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 106     real             *ewtab;
 107     __m128           dummy_mask,cutoff_mask;
 108     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 109     __m128           one     = _mm_set1_ps(1.0);
 110     __m128           two     = _mm_set1_ps(2.0);
 111     x                = xx[0];
 112     f                = ff[0];
 113
 114     nri              = nlist->nri;
 115     iinr             = nlist->iinr;
 116     jindex           = nlist->jindex;
 117     jjnr             = nlist->jjnr;
 118     shiftidx         = nlist->shift;
 119     gid              = nlist->gid;
 120     shiftvec         = fr->shift_vec[0];
 121     fshift           = fr->fshift[0];
 122     facel            = _mm_set1_ps(fr->epsfac);
 123     charge           = mdatoms->chargeA;
 124     nvdwtype         = fr->ntype;
 125     vdwparam         = fr->nbfp;
 126     vdwtype          = mdatoms->typeA;
 127     vdwgridparam     = fr->ljpme_c6grid;
 128     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
 129     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
 130     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
 131
 132     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 133     beta             = _mm_set1_ps(fr->ic->ewaldcoeff_q);
 134     beta2            = _mm_mul_ps(beta,beta);
 135     beta3            = _mm_mul_ps(beta,beta2);
 136     ewtab            = fr->ic->tabq_coul_FDV0;
 137     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 138     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 139
 140     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 141     rcutoff_scalar   = fr->rcoulomb;
 142     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 143     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 144
 145     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
 146     rvdw             = _mm_set1_ps(fr->rvdw);
 147
 148     /* Avoid stupid compiler warnings */
 149     jnrA = jnrB = jnrC = jnrD = 0;
 150     j_coord_offsetA = 0;
 151     j_coord_offsetB = 0;
 152     j_coord_offsetC = 0;
 153     j_coord_offsetD = 0;
 154
 155     outeriter        = 0;
 156     inneriter        = 0;
 157
 158     for(iidx=0;iidx<4*DIM;iidx++)
 159     {
 160         scratch[iidx] = 0.0;
 161     }
 162
 163     /* Start outer loop over neighborlists */
 164     for(iidx=0; iidx<nri; iidx++)
 165     {
 166         /* Load shift vector for this list */
 167         i_shift_offset   = DIM*shiftidx[iidx];
 168
 169         /* Load limits for loop over neighbors */
 170         j_index_start    = jindex[iidx];
 171         j_index_end      = jindex[iidx+1];
 172
 173         /* Get outer coordinate index */
 174         inr              = iinr[iidx];
 175         i_coord_offset   = DIM*inr;
 176
 177         /* Load i particle coords and add shift vector */
 178         gmx_mm_load_shift_and_1rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 179
 180         fix0             = _mm_setzero_ps();
 181         fiy0             = _mm_setzero_ps();
 182         fiz0             = _mm_setzero_ps();
 183
 184         /* Load parameters for i particles */
 185         iq0              = _mm_mul_ps(facel,_mm_load1_ps(charge+inr+0));
 186         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 187
 188         /* Reset potential sums */
 189         velecsum         = _mm_setzero_ps();
 190         vvdwsum          = _mm_setzero_ps();
 191
 192         /* Start inner kernel loop */
 193         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 194         {
 195
 196             /* Get j neighbor index, and coordinate index */
 197             jnrA             = jjnr[jidx];
 198             jnrB             = jjnr[jidx+1];
 199             jnrC             = jjnr[jidx+2];
 200             jnrD             = jjnr[jidx+3];
 201             j_coord_offsetA  = DIM*jnrA;
 202             j_coord_offsetB  = DIM*jnrB;
 203             j_coord_offsetC  = DIM*jnrC;
 204             j_coord_offsetD  = DIM*jnrD;
 205
 206             /* load j atom coordinates */
 207             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 208                                               x+j_coord_offsetC,x+j_coord_offsetD,
 209                                               &jx0,&jy0,&jz0);
 210
 211             /* Calculate displacement vector */
 212             dx00             = _mm_sub_ps(ix0,jx0);
 213             dy00             = _mm_sub_ps(iy0,jy0);
 214             dz00             = _mm_sub_ps(iz0,jz0);
 215
 216             /* Calculate squared distance and things based on it */
 217             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 218
 219             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 220
 221             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 222
 223             /* Load parameters for j particles */
 224             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 225                                                               charge+jnrC+0,charge+jnrD+0);
 226             vdwjidx0A        = 2*vdwtype[jnrA+0];
 227             vdwjidx0B        = 2*vdwtype[jnrB+0];
 228             vdwjidx0C        = 2*vdwtype[jnrC+0];
 229             vdwjidx0D        = 2*vdwtype[jnrD+0];
 230
 231             /**************************
 232              * CALCULATE INTERACTIONS *
 233              **************************/
 234
 235             if (gmx_mm_any_lt(rsq00,rcutoff2))
 236             {
 237
 238             r00              = _mm_mul_ps(rsq00,rinv00);
 239
 240             /* Compute parameters for interactions between i and j atoms */
 241             qq00             = _mm_mul_ps(iq0,jq0);
 242             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 243                                          vdwparam+vdwioffset0+vdwjidx0B,
 244                                          vdwparam+vdwioffset0+vdwjidx0C,
 245                                          vdwparam+vdwioffset0+vdwjidx0D,
 246                                          &c6_00,&c12_00);
 247
 248             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 249                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 250                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 251                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 252
 253             /* EWALD ELECTROSTATICS */
 254
 255             /* Analytical PME correction */
 256             zeta2            = _mm_mul_ps(beta2,rsq00);
 257             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 258             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 259             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 260             felec            = _mm_mul_ps(qq00,felec);
 261             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 262             velec            = _mm_nmacc_ps(pmecorrV,beta,_mm_sub_ps(rinv00,sh_ewald));
 263             velec            = _mm_mul_ps(qq00,velec);
 264
 265             /* Analytical LJ-PME */
 266             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 267             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 268             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 269             exponent         = gmx_simd_exp_r(ewcljrsq);
 270             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 271             poly             = _mm_mul_ps(exponent,_mm_macc_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half,_mm_sub_ps(one,ewcljrsq)));
 272             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 273             vvdw6            = _mm_mul_ps(_mm_macc_ps(-c6grid_00,_mm_sub_ps(one,poly),c6_00),rinvsix);
 274             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 275             vvdw             = _mm_msub_ps(_mm_nmacc_ps(c12_00,_mm_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6),vvdw12),one_twelfth,
 276                                           _mm_mul_ps(_mm_sub_ps(vvdw6,_mm_macc_ps(c6grid_00,sh_lj_ewald,_mm_mul_ps(c6_00,sh_vdw_invrcut6))),one_sixth));
 277             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 278             fvdw             = _mm_mul_ps(_mm_add_ps(vvdw12,_mm_msub_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6),vvdw6)),rinvsq00);
 279
 280             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 281
 282             /* Update potential sum for this i atom from the interaction with this j atom. */
 283             velec            = _mm_and_ps(velec,cutoff_mask);
 284             velecsum         = _mm_add_ps(velecsum,velec);
 285             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 286             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 287
 288             fscal            = _mm_add_ps(felec,fvdw);
 289
 290             fscal            = _mm_and_ps(fscal,cutoff_mask);
 291
 292              /* Update vectorial force */
 293             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 294             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 295             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 296
 297             fjptrA             = f+j_coord_offsetA;
 298             fjptrB             = f+j_coord_offsetB;
 299             fjptrC             = f+j_coord_offsetC;
 300             fjptrD             = f+j_coord_offsetD;
 301             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 302                                                    _mm_mul_ps(dx00,fscal),
 303                                                    _mm_mul_ps(dy00,fscal),
 304                                                    _mm_mul_ps(dz00,fscal));
 305
 306             }
 307
 308             /* Inner loop uses 63 flops */
 309         }
 310
 311         if(jidx<j_index_end)
 312         {
 313
 314             /* Get j neighbor index, and coordinate index */
 315             jnrlistA         = jjnr[jidx];
 316             jnrlistB         = jjnr[jidx+1];
 317             jnrlistC         = jjnr[jidx+2];
 318             jnrlistD         = jjnr[jidx+3];
 319             /* Sign of each element will be negative for non-real atoms.
 320              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 321              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 322              */
 323             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 324             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 325             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 326             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 327             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 328             j_coord_offsetA  = DIM*jnrA;
 329             j_coord_offsetB  = DIM*jnrB;
 330             j_coord_offsetC  = DIM*jnrC;
 331             j_coord_offsetD  = DIM*jnrD;
 332
 333             /* load j atom coordinates */
 334             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 335                                               x+j_coord_offsetC,x+j_coord_offsetD,
 336                                               &jx0,&jy0,&jz0);
 337
 338             /* Calculate displacement vector */
 339             dx00             = _mm_sub_ps(ix0,jx0);
 340             dy00             = _mm_sub_ps(iy0,jy0);
 341             dz00             = _mm_sub_ps(iz0,jz0);
 342
 343             /* Calculate squared distance and things based on it */
 344             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 345
 346             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 347
 348             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 349
 350             /* Load parameters for j particles */
 351             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 352                                                               charge+jnrC+0,charge+jnrD+0);
 353             vdwjidx0A        = 2*vdwtype[jnrA+0];
 354             vdwjidx0B        = 2*vdwtype[jnrB+0];
 355             vdwjidx0C        = 2*vdwtype[jnrC+0];
 356             vdwjidx0D        = 2*vdwtype[jnrD+0];
 357
 358             /**************************
 359              * CALCULATE INTERACTIONS *
 360              **************************/
 361
 362             if (gmx_mm_any_lt(rsq00,rcutoff2))
 363             {
 364
 365             r00              = _mm_mul_ps(rsq00,rinv00);
 366             r00              = _mm_andnot_ps(dummy_mask,r00);
 367
 368             /* Compute parameters for interactions between i and j atoms */
 369             qq00             = _mm_mul_ps(iq0,jq0);
 370             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 371                                          vdwparam+vdwioffset0+vdwjidx0B,
 372                                          vdwparam+vdwioffset0+vdwjidx0C,
 373                                          vdwparam+vdwioffset0+vdwjidx0D,
 374                                          &c6_00,&c12_00);
 375
 376             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 377                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 378                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 379                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 380
 381             /* EWALD ELECTROSTATICS */
 382
 383             /* Analytical PME correction */
 384             zeta2            = _mm_mul_ps(beta2,rsq00);
 385             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 386             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 387             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 388             felec            = _mm_mul_ps(qq00,felec);
 389             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 390             velec            = _mm_nmacc_ps(pmecorrV,beta,_mm_sub_ps(rinv00,sh_ewald));
 391             velec            = _mm_mul_ps(qq00,velec);
 392
 393             /* Analytical LJ-PME */
 394             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 395             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 396             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 397             exponent         = gmx_simd_exp_r(ewcljrsq);
 398             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 399             poly             = _mm_mul_ps(exponent,_mm_macc_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half,_mm_sub_ps(one,ewcljrsq)));
 400             /* vvdw6 = [C6 - C6grid * (1-poly)]/r6 */
 401             vvdw6            = _mm_mul_ps(_mm_macc_ps(-c6grid_00,_mm_sub_ps(one,poly),c6_00),rinvsix);
 402             vvdw12           = _mm_mul_ps(c12_00,_mm_mul_ps(rinvsix,rinvsix));
 403             vvdw             = _mm_msub_ps(_mm_nmacc_ps(c12_00,_mm_mul_ps(sh_vdw_invrcut6,sh_vdw_invrcut6),vvdw12),one_twelfth,
 404                                           _mm_mul_ps(_mm_sub_ps(vvdw6,_mm_macc_ps(c6grid_00,sh_lj_ewald,_mm_mul_ps(c6_00,sh_vdw_invrcut6))),one_sixth));
 405             /* fvdw = vvdw12/r - (vvdw6/r + (C6grid * exponent * beta^6)/r) */
 406             fvdw             = _mm_mul_ps(_mm_add_ps(vvdw12,_mm_msub_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6),vvdw6)),rinvsq00);
 407
 408             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 409
 410             /* Update potential sum for this i atom from the interaction with this j atom. */
 411             velec            = _mm_and_ps(velec,cutoff_mask);
 412             velec            = _mm_andnot_ps(dummy_mask,velec);
 413             velecsum         = _mm_add_ps(velecsum,velec);
 414             vvdw             = _mm_and_ps(vvdw,cutoff_mask);
 415             vvdw             = _mm_andnot_ps(dummy_mask,vvdw);
 416             vvdwsum          = _mm_add_ps(vvdwsum,vvdw);
 417
 418             fscal            = _mm_add_ps(felec,fvdw);
 419
 420             fscal            = _mm_and_ps(fscal,cutoff_mask);
 421
 422             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 423
 424              /* Update vectorial force */
 425             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 426             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 427             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 428
 429             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 430             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 431             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 432             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 433             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 434                                                    _mm_mul_ps(dx00,fscal),
 435                                                    _mm_mul_ps(dy00,fscal),
 436                                                    _mm_mul_ps(dz00,fscal));
 437
 438             }
 439
 440             /* Inner loop uses 64 flops */
 441         }
 442
 443         /* End of innermost loop */
 444
 445         gmx_mm_update_iforce_1atom_swizzle_ps(fix0,fiy0,fiz0,
 446                                               f+i_coord_offset,fshift+i_shift_offset);
 447
 448         ggid                        = gid[iidx];
 449         /* Update potential energies */
 450         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 451         gmx_mm_update_1pot_ps(vvdwsum,kernel_data->energygrp_vdw+ggid);
 452
 453         /* Increment number of inner iterations */
 454         inneriter                  += j_index_end - j_index_start;
 455
 456         /* Outer loop uses 9 flops */
 457     }
 458
 459     /* Increment number of outer iterations */
 460     outeriter        += nri;
 461
 462     /* Update outer/inner flops */
 463
 464     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_VF,outeriter*9 + inneriter*64);
 465 }
 466 /*
 467  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_F_avx_128_fma_single
 468  * Electrostatics interaction: Ewald
 469  * VdW interaction:            LJEwald
 470  * Geometry:                   Particle-Particle
 471  * Calculate force/pot:        Force
 472  */
 473 void
 474 nb_kernel_ElecEwSh_VdwLJEwSh_GeomP1P1_F_avx_128_fma_single
 475                     (t_nblist                    * gmx_restrict       nlist,
 476                      rvec                        * gmx_restrict          xx,
 477                      rvec                        * gmx_restrict          ff,
 478                      t_forcerec                  * gmx_restrict          fr,
 479                      t_mdatoms                   * gmx_restrict     mdatoms,
 480                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 481                      t_nrnb                      * gmx_restrict        nrnb)
 482 {
 483     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 484      * just 0 for non-waters.
 485      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
 486      * jnr indices corresponding to data put in the four positions in the SIMD register.
 487      */
 488     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 489     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 490     int              jnrA,jnrB,jnrC,jnrD;
 491     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 492     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 493     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 494     real             rcutoff_scalar;
 495     real             *shiftvec,*fshift,*x,*f;
 496     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 497     real             scratch[4*DIM];
 498     __m128           fscal,rcutoff,rcutoff2,jidxall;
 499     int              vdwioffset0;
 500     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 501     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 502     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 503     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 504     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 505     real             *charge;
 506     int              nvdwtype;
 507     __m128           rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 508     int              *vdwtype;
 509     real             *vdwparam;
 510     __m128           one_sixth   = _mm_set1_ps(1.0/6.0);
 511     __m128           one_twelfth = _mm_set1_ps(1.0/12.0);
 512     __m128           c6grid_00;
 513     real             *vdwgridparam;
 514     __m128           ewclj,ewclj2,ewclj6,ewcljrsq,poly,exponent,f6A,f6B,sh_lj_ewald;
 515     __m128           one_half = _mm_set1_ps(0.5);
 516     __m128           minus_one = _mm_set1_ps(-1.0);
 517     __m128i          ewitab;
 518     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 519     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 520     real             *ewtab;
 521     __m128           dummy_mask,cutoff_mask;
 522     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 523     __m128           one     = _mm_set1_ps(1.0);
 524     __m128           two     = _mm_set1_ps(2.0);
 525     x                = xx[0];
 526     f                = ff[0];
 527
 528     nri              = nlist->nri;
 529     iinr             = nlist->iinr;
 530     jindex           = nlist->jindex;
 531     jjnr             = nlist->jjnr;
 532     shiftidx         = nlist->shift;
 533     gid              = nlist->gid;
 534     shiftvec         = fr->shift_vec[0];
 535     fshift           = fr->fshift[0];
 536     facel            = _mm_set1_ps(fr->epsfac);
 537     charge           = mdatoms->chargeA;
 538     nvdwtype         = fr->ntype;
 539     vdwparam         = fr->nbfp;
 540     vdwtype          = mdatoms->typeA;
 541     vdwgridparam     = fr->ljpme_c6grid;
 542     sh_lj_ewald      = _mm_set1_ps(fr->ic->sh_lj_ewald);
 543     ewclj            = _mm_set1_ps(fr->ewaldcoeff_lj);
 544     ewclj2           = _mm_mul_ps(minus_one,_mm_mul_ps(ewclj,ewclj));
 545
 546     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 547     beta             = _mm_set1_ps(fr->ic->ewaldcoeff_q);
 548     beta2            = _mm_mul_ps(beta,beta);
 549     beta3            = _mm_mul_ps(beta,beta2);
 550     ewtab            = fr->ic->tabq_coul_F;
 551     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 552     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 553
 554     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 555     rcutoff_scalar   = fr->rcoulomb;
 556     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 557     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 558
 559     sh_vdw_invrcut6  = _mm_set1_ps(fr->ic->sh_invrc6);
 560     rvdw             = _mm_set1_ps(fr->rvdw);
 561
 562     /* Avoid stupid compiler warnings */
 563     jnrA = jnrB = jnrC = jnrD = 0;
 564     j_coord_offsetA = 0;
 565     j_coord_offsetB = 0;
 566     j_coord_offsetC = 0;
 567     j_coord_offsetD = 0;
 568
 569     outeriter        = 0;
 570     inneriter        = 0;
 571
 572     for(iidx=0;iidx<4*DIM;iidx++)
 573     {
 574         scratch[iidx] = 0.0;
 575     }
 576
 577     /* Start outer loop over neighborlists */
 578     for(iidx=0; iidx<nri; iidx++)
 579     {
 580         /* Load shift vector for this list */
 581         i_shift_offset   = DIM*shiftidx[iidx];
 582
 583         /* Load limits for loop over neighbors */
 584         j_index_start    = jindex[iidx];
 585         j_index_end      = jindex[iidx+1];
 586
 587         /* Get outer coordinate index */
 588         inr              = iinr[iidx];
 589         i_coord_offset   = DIM*inr;
 590
 591         /* Load i particle coords and add shift vector */
 592         gmx_mm_load_shift_and_1rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,&ix0,&iy0,&iz0);
 593
 594         fix0             = _mm_setzero_ps();
 595         fiy0             = _mm_setzero_ps();
 596         fiz0             = _mm_setzero_ps();
 597
 598         /* Load parameters for i particles */
 599         iq0              = _mm_mul_ps(facel,_mm_load1_ps(charge+inr+0));
 600         vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 601
 602         /* Start inner kernel loop */
 603         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 604         {
 605
 606             /* Get j neighbor index, and coordinate index */
 607             jnrA             = jjnr[jidx];
 608             jnrB             = jjnr[jidx+1];
 609             jnrC             = jjnr[jidx+2];
 610             jnrD             = jjnr[jidx+3];
 611             j_coord_offsetA  = DIM*jnrA;
 612             j_coord_offsetB  = DIM*jnrB;
 613             j_coord_offsetC  = DIM*jnrC;
 614             j_coord_offsetD  = DIM*jnrD;
 615
 616             /* load j atom coordinates */
 617             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 618                                               x+j_coord_offsetC,x+j_coord_offsetD,
 619                                               &jx0,&jy0,&jz0);
 620
 621             /* Calculate displacement vector */
 622             dx00             = _mm_sub_ps(ix0,jx0);
 623             dy00             = _mm_sub_ps(iy0,jy0);
 624             dz00             = _mm_sub_ps(iz0,jz0);
 625
 626             /* Calculate squared distance and things based on it */
 627             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 628
 629             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 630
 631             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 632
 633             /* Load parameters for j particles */
 634             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 635                                                               charge+jnrC+0,charge+jnrD+0);
 636             vdwjidx0A        = 2*vdwtype[jnrA+0];
 637             vdwjidx0B        = 2*vdwtype[jnrB+0];
 638             vdwjidx0C        = 2*vdwtype[jnrC+0];
 639             vdwjidx0D        = 2*vdwtype[jnrD+0];
 640
 641             /**************************
 642              * CALCULATE INTERACTIONS *
 643              **************************/
 644
 645             if (gmx_mm_any_lt(rsq00,rcutoff2))
 646             {
 647
 648             r00              = _mm_mul_ps(rsq00,rinv00);
 649
 650             /* Compute parameters for interactions between i and j atoms */
 651             qq00             = _mm_mul_ps(iq0,jq0);
 652             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 653                                          vdwparam+vdwioffset0+vdwjidx0B,
 654                                          vdwparam+vdwioffset0+vdwjidx0C,
 655                                          vdwparam+vdwioffset0+vdwjidx0D,
 656                                          &c6_00,&c12_00);
 657
 658             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 659                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 660                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 661                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 662
 663             /* EWALD ELECTROSTATICS */
 664
 665             /* Analytical PME correction */
 666             zeta2            = _mm_mul_ps(beta2,rsq00);
 667             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 668             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 669             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 670             felec            = _mm_mul_ps(qq00,felec);
 671
 672             /* Analytical LJ-PME */
 673             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 674             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 675             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 676             exponent         = gmx_simd_exp_r(ewcljrsq);
 677             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 678             poly             = _mm_mul_ps(exponent,_mm_macc_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half,_mm_sub_ps(one,ewcljrsq)));
 679             /* f6A = 6 * C6grid * (1 - poly) */
 680             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
 681             /* f6B = C6grid * exponent * beta^6 */
 682             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
 683             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 684             fvdw              = _mm_mul_ps(_mm_macc_ps(_mm_msub_ps(c12_00,rinvsix,_mm_sub_ps(c6_00,f6A)),rinvsix,f6B),rinvsq00);
 685
 686             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 687
 688             fscal            = _mm_add_ps(felec,fvdw);
 689
 690             fscal            = _mm_and_ps(fscal,cutoff_mask);
 691
 692              /* Update vectorial force */
 693             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 694             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 695             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 696
 697             fjptrA             = f+j_coord_offsetA;
 698             fjptrB             = f+j_coord_offsetB;
 699             fjptrC             = f+j_coord_offsetC;
 700             fjptrD             = f+j_coord_offsetD;
 701             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 702                                                    _mm_mul_ps(dx00,fscal),
 703                                                    _mm_mul_ps(dy00,fscal),
 704                                                    _mm_mul_ps(dz00,fscal));
 705
 706             }
 707
 708             /* Inner loop uses 52 flops */
 709         }
 710
 711         if(jidx<j_index_end)
 712         {
 713
 714             /* Get j neighbor index, and coordinate index */
 715             jnrlistA         = jjnr[jidx];
 716             jnrlistB         = jjnr[jidx+1];
 717             jnrlistC         = jjnr[jidx+2];
 718             jnrlistD         = jjnr[jidx+3];
 719             /* Sign of each element will be negative for non-real atoms.
 720              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 721              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 722              */
 723             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 724             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 725             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 726             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 727             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 728             j_coord_offsetA  = DIM*jnrA;
 729             j_coord_offsetB  = DIM*jnrB;
 730             j_coord_offsetC  = DIM*jnrC;
 731             j_coord_offsetD  = DIM*jnrD;
 732
 733             /* load j atom coordinates */
 734             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 735                                               x+j_coord_offsetC,x+j_coord_offsetD,
 736                                               &jx0,&jy0,&jz0);
 737
 738             /* Calculate displacement vector */
 739             dx00             = _mm_sub_ps(ix0,jx0);
 740             dy00             = _mm_sub_ps(iy0,jy0);
 741             dz00             = _mm_sub_ps(iz0,jz0);
 742
 743             /* Calculate squared distance and things based on it */
 744             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 745
 746             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 747
 748             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 749
 750             /* Load parameters for j particles */
 751             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 752                                                               charge+jnrC+0,charge+jnrD+0);
 753             vdwjidx0A        = 2*vdwtype[jnrA+0];
 754             vdwjidx0B        = 2*vdwtype[jnrB+0];
 755             vdwjidx0C        = 2*vdwtype[jnrC+0];
 756             vdwjidx0D        = 2*vdwtype[jnrD+0];
 757
 758             /**************************
 759              * CALCULATE INTERACTIONS *
 760              **************************/
 761
 762             if (gmx_mm_any_lt(rsq00,rcutoff2))
 763             {
 764
 765             r00              = _mm_mul_ps(rsq00,rinv00);
 766             r00              = _mm_andnot_ps(dummy_mask,r00);
 767
 768             /* Compute parameters for interactions between i and j atoms */
 769             qq00             = _mm_mul_ps(iq0,jq0);
 770             gmx_mm_load_4pair_swizzle_ps(vdwparam+vdwioffset0+vdwjidx0A,
 771                                          vdwparam+vdwioffset0+vdwjidx0B,
 772                                          vdwparam+vdwioffset0+vdwjidx0C,
 773                                          vdwparam+vdwioffset0+vdwjidx0D,
 774                                          &c6_00,&c12_00);
 775
 776             c6grid_00       = gmx_mm_load_4real_swizzle_ps(vdwgridparam+vdwioffset0+vdwjidx0A,
 777                                                                vdwgridparam+vdwioffset0+vdwjidx0B,
 778                                                                vdwgridparam+vdwioffset0+vdwjidx0C,
 779                                                                vdwgridparam+vdwioffset0+vdwjidx0D);
 780
 781             /* EWALD ELECTROSTATICS */
 782
 783             /* Analytical PME correction */
 784             zeta2            = _mm_mul_ps(beta2,rsq00);
 785             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 786             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 787             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 788             felec            = _mm_mul_ps(qq00,felec);
 789
 790             /* Analytical LJ-PME */
 791             rinvsix          = _mm_mul_ps(_mm_mul_ps(rinvsq00,rinvsq00),rinvsq00);
 792             ewcljrsq         = _mm_mul_ps(ewclj2,rsq00);
 793             ewclj6           = _mm_mul_ps(ewclj2,_mm_mul_ps(ewclj2,ewclj2));
 794             exponent         = gmx_simd_exp_r(ewcljrsq);
 795             /* poly = exp(-(beta*r)^2) * (1 + (beta*r)^2 + (beta*r)^4 /2) */
 796             poly             = _mm_mul_ps(exponent,_mm_macc_ps(_mm_mul_ps(ewcljrsq,ewcljrsq),one_half,_mm_sub_ps(one,ewcljrsq)));
 797             /* f6A = 6 * C6grid * (1 - poly) */
 798             f6A              = _mm_mul_ps(c6grid_00,_mm_sub_ps(one,poly));
 799             /* f6B = C6grid * exponent * beta^6 */
 800             f6B              = _mm_mul_ps(_mm_mul_ps(c6grid_00,one_sixth),_mm_mul_ps(exponent,ewclj6));
 801             /* fvdw = 12*C12/r13 - ((6*C6 - f6A)/r6 + f6B)/r */
 802             fvdw              = _mm_mul_ps(_mm_macc_ps(_mm_msub_ps(c12_00,rinvsix,_mm_sub_ps(c6_00,f6A)),rinvsix,f6B),rinvsq00);
 803
 804             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 805
 806             fscal            = _mm_add_ps(felec,fvdw);
 807
 808             fscal            = _mm_and_ps(fscal,cutoff_mask);
 809
 810             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 811
 812              /* Update vectorial force */
 813             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 814             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 815             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 816
 817             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 818             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 819             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 820             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 821             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,
 822                                                    _mm_mul_ps(dx00,fscal),
 823                                                    _mm_mul_ps(dy00,fscal),
 824                                                    _mm_mul_ps(dz00,fscal));
 825
 826             }
 827
 828             /* Inner loop uses 53 flops */
 829         }
 830
 831         /* End of innermost loop */
 832
 833         gmx_mm_update_iforce_1atom_swizzle_ps(fix0,fiy0,fiz0,
 834                                               f+i_coord_offset,fshift+i_shift_offset);
 835
 836         /* Increment number of inner iterations */
 837         inneriter                  += j_index_end - j_index_start;
 838
 839         /* Outer loop uses 7 flops */
 840     }
 841
 842     /* Increment number of outer iterations */
 843     outeriter        += nri;
 844
 845     /* Update outer/inner flops */
 846
 847     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_F,outeriter*7 + inneriter*53);
 848 }