src/gromacs/gmxlib/nonbonded/nb_kernel_avx_128_fma_double/nb_kernel_ElecCSTab_VdwLJ_GeomW3P1_avx_128_fma_double.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_double kernel generator.
  37  */
  38 #include "config.h"
  39
  40 #include <math.h>
  41
  42 #include "../nb_kernel.h"
  43 #include "gromacs/legacyheaders/types/simple.h"
  44 #include "gromacs/math/vec.h"
  45 #include "gromacs/legacyheaders/nrnb.h"
  46
  47 #include "gromacs/simd/math_x86_avx_128_fma_double.h"
  48 #include "kernelutil_x86_avx_128_fma_double.h"
  49
  50 /*
  51  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwLJ_GeomW3P1_VF_avx_128_fma_double
  52  * Electrostatics interaction: CubicSplineTable
  53  * VdW interaction:            LennardJones
  54  * Geometry:                   Water3-Particle
  55  * Calculate force/pot:        PotentialAndForce
  56  */
  57 void
  58 nb_kernel_ElecCSTab_VdwLJ_GeomW3P1_VF_avx_128_fma_double
  59                     (t_nblist                    * gmx_restrict       nlist,
  60                      rvec                        * gmx_restrict          xx,
  61                      rvec                        * gmx_restrict          ff,
  62                      t_forcerec                  * gmx_restrict          fr,
  63                      t_mdatoms                   * gmx_restrict     mdatoms,
  64                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
  65                      t_nrnb                      * gmx_restrict        nrnb)
  66 {
  67     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  68      * just 0 for non-waters.
  69      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
  70      * jnr indices corresponding to data put in the four positions in the SIMD register.
  71      */
  72     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  73     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  74     int              jnrA,jnrB;
  75     int              j_coord_offsetA,j_coord_offsetB;
  76     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  77     real             rcutoff_scalar;
  78     real             *shiftvec,*fshift,*x,*f;
  79     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
  80     int              vdwioffset0;
  81     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  82     int              vdwioffset1;
  83     __m128d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  84     int              vdwioffset2;
  85     __m128d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  86     int              vdwjidx0A,vdwjidx0B;
  87     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  88     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  89     __m128d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  90     __m128d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  91     __m128d          velec,felec,velecsum,facel,crf,krf,krf2;
  92     real             *charge;
  93     int              nvdwtype;
  94     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
  95     int              *vdwtype;
  96     real             *vdwparam;
  97     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
  98     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
  99     __m128i          vfitab;
 100     __m128i          ifour       = _mm_set1_epi32(4);
 101     __m128d          rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 102     real             *vftab;
 103     __m128d          dummy_mask,cutoff_mask;
 104     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
 105     __m128d          one     = _mm_set1_pd(1.0);
 106     __m128d          two     = _mm_set1_pd(2.0);
 107     x                = xx[0];
 108     f                = ff[0];
 109
 110     nri              = nlist->nri;
 111     iinr             = nlist->iinr;
 112     jindex           = nlist->jindex;
 113     jjnr             = nlist->jjnr;
 114     shiftidx         = nlist->shift;
 115     gid              = nlist->gid;
 116     shiftvec         = fr->shift_vec[0];
 117     fshift           = fr->fshift[0];
 118     facel            = _mm_set1_pd(fr->epsfac);
 119     charge           = mdatoms->chargeA;
 120     nvdwtype         = fr->ntype;
 121     vdwparam         = fr->nbfp;
 122     vdwtype          = mdatoms->typeA;
 123
 124     vftab            = kernel_data->table_elec->data;
 125     vftabscale       = _mm_set1_pd(kernel_data->table_elec->scale);
 126
 127     /* Setup water-specific parameters */
 128     inr              = nlist->iinr[0];
 129     iq0              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
 130     iq1              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
 131     iq2              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
 132     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 133
 134     /* Avoid stupid compiler warnings */
 135     jnrA = jnrB = 0;
 136     j_coord_offsetA = 0;
 137     j_coord_offsetB = 0;
 138
 139     outeriter        = 0;
 140     inneriter        = 0;
 141
 142     /* Start outer loop over neighborlists */
 143     for(iidx=0; iidx<nri; iidx++)
 144     {
 145         /* Load shift vector for this list */
 146         i_shift_offset   = DIM*shiftidx[iidx];
 147
 148         /* Load limits for loop over neighbors */
 149         j_index_start    = jindex[iidx];
 150         j_index_end      = jindex[iidx+1];
 151
 152         /* Get outer coordinate index */
 153         inr              = iinr[iidx];
 154         i_coord_offset   = DIM*inr;
 155
 156         /* Load i particle coords and add shift vector */
 157         gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 158                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 159
 160         fix0             = _mm_setzero_pd();
 161         fiy0             = _mm_setzero_pd();
 162         fiz0             = _mm_setzero_pd();
 163         fix1             = _mm_setzero_pd();
 164         fiy1             = _mm_setzero_pd();
 165         fiz1             = _mm_setzero_pd();
 166         fix2             = _mm_setzero_pd();
 167         fiy2             = _mm_setzero_pd();
 168         fiz2             = _mm_setzero_pd();
 169
 170         /* Reset potential sums */
 171         velecsum         = _mm_setzero_pd();
 172         vvdwsum          = _mm_setzero_pd();
 173
 174         /* Start inner kernel loop */
 175         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 176         {
 177
 178             /* Get j neighbor index, and coordinate index */
 179             jnrA             = jjnr[jidx];
 180             jnrB             = jjnr[jidx+1];
 181             j_coord_offsetA  = DIM*jnrA;
 182             j_coord_offsetB  = DIM*jnrB;
 183
 184             /* load j atom coordinates */
 185             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 186                                               &jx0,&jy0,&jz0);
 187
 188             /* Calculate displacement vector */
 189             dx00             = _mm_sub_pd(ix0,jx0);
 190             dy00             = _mm_sub_pd(iy0,jy0);
 191             dz00             = _mm_sub_pd(iz0,jz0);
 192             dx10             = _mm_sub_pd(ix1,jx0);
 193             dy10             = _mm_sub_pd(iy1,jy0);
 194             dz10             = _mm_sub_pd(iz1,jz0);
 195             dx20             = _mm_sub_pd(ix2,jx0);
 196             dy20             = _mm_sub_pd(iy2,jy0);
 197             dz20             = _mm_sub_pd(iz2,jz0);
 198
 199             /* Calculate squared distance and things based on it */
 200             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 201             rsq10            = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
 202             rsq20            = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
 203
 204             rinv00           = gmx_mm_invsqrt_pd(rsq00);
 205             rinv10           = gmx_mm_invsqrt_pd(rsq10);
 206             rinv20           = gmx_mm_invsqrt_pd(rsq20);
 207
 208             rinvsq00         = _mm_mul_pd(rinv00,rinv00);
 209
 210             /* Load parameters for j particles */
 211             jq0              = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
 212             vdwjidx0A        = 2*vdwtype[jnrA+0];
 213             vdwjidx0B        = 2*vdwtype[jnrB+0];
 214
 215             fjx0             = _mm_setzero_pd();
 216             fjy0             = _mm_setzero_pd();
 217             fjz0             = _mm_setzero_pd();
 218
 219             /**************************
 220              * CALCULATE INTERACTIONS *
 221              **************************/
 222
 223             r00              = _mm_mul_pd(rsq00,rinv00);
 224
 225             /* Compute parameters for interactions between i and j atoms */
 226             qq00             = _mm_mul_pd(iq0,jq0);
 227             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 228                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 229
 230             /* Calculate table index by multiplying r with table scale and truncate to integer */
 231             rt               = _mm_mul_pd(r00,vftabscale);
 232             vfitab           = _mm_cvttpd_epi32(rt);
 233 #ifdef __XOP__
 234             vfeps            = _mm_frcz_pd(rt);
 235 #else
 236             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 237 #endif
 238             twovfeps         = _mm_add_pd(vfeps,vfeps);
 239             vfitab           = _mm_slli_epi32(vfitab,2);
 240
 241             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 242             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 243             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 244             GMX_MM_TRANSPOSE2_PD(Y,F);
 245             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 246             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 247             GMX_MM_TRANSPOSE2_PD(G,H);
 248             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 249             VV               = _mm_macc_pd(vfeps,Fp,Y);
 250             velec            = _mm_mul_pd(qq00,VV);
 251             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 252             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq00,FF),_mm_mul_pd(vftabscale,rinv00)));
 253
 254             /* LENNARD-JONES DISPERSION/REPULSION */
 255
 256             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 257             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 258             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 259             vvdw             = _mm_msub_pd( vvdw12,one_twelfth, _mm_mul_pd(vvdw6,one_sixth) );
 260             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 261
 262             /* Update potential sum for this i atom from the interaction with this j atom. */
 263             velecsum         = _mm_add_pd(velecsum,velec);
 264             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 265
 266             fscal            = _mm_add_pd(felec,fvdw);
 267
 268             /* Update vectorial force */
 269             fix0             = _mm_macc_pd(dx00,fscal,fix0);
 270             fiy0             = _mm_macc_pd(dy00,fscal,fiy0);
 271             fiz0             = _mm_macc_pd(dz00,fscal,fiz0);
 272
 273             fjx0             = _mm_macc_pd(dx00,fscal,fjx0);
 274             fjy0             = _mm_macc_pd(dy00,fscal,fjy0);
 275             fjz0             = _mm_macc_pd(dz00,fscal,fjz0);
 276
 277             /**************************
 278              * CALCULATE INTERACTIONS *
 279              **************************/
 280
 281             r10              = _mm_mul_pd(rsq10,rinv10);
 282
 283             /* Compute parameters for interactions between i and j atoms */
 284             qq10             = _mm_mul_pd(iq1,jq0);
 285
 286             /* Calculate table index by multiplying r with table scale and truncate to integer */
 287             rt               = _mm_mul_pd(r10,vftabscale);
 288             vfitab           = _mm_cvttpd_epi32(rt);
 289 #ifdef __XOP__
 290             vfeps            = _mm_frcz_pd(rt);
 291 #else
 292             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 293 #endif
 294             twovfeps         = _mm_add_pd(vfeps,vfeps);
 295             vfitab           = _mm_slli_epi32(vfitab,2);
 296
 297             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 298             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 299             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 300             GMX_MM_TRANSPOSE2_PD(Y,F);
 301             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 302             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 303             GMX_MM_TRANSPOSE2_PD(G,H);
 304             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 305             VV               = _mm_macc_pd(vfeps,Fp,Y);
 306             velec            = _mm_mul_pd(qq10,VV);
 307             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 308             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq10,FF),_mm_mul_pd(vftabscale,rinv10)));
 309
 310             /* Update potential sum for this i atom from the interaction with this j atom. */
 311             velecsum         = _mm_add_pd(velecsum,velec);
 312
 313             fscal            = felec;
 314
 315             /* Update vectorial force */
 316             fix1             = _mm_macc_pd(dx10,fscal,fix1);
 317             fiy1             = _mm_macc_pd(dy10,fscal,fiy1);
 318             fiz1             = _mm_macc_pd(dz10,fscal,fiz1);
 319
 320             fjx0             = _mm_macc_pd(dx10,fscal,fjx0);
 321             fjy0             = _mm_macc_pd(dy10,fscal,fjy0);
 322             fjz0             = _mm_macc_pd(dz10,fscal,fjz0);
 323
 324             /**************************
 325              * CALCULATE INTERACTIONS *
 326              **************************/
 327
 328             r20              = _mm_mul_pd(rsq20,rinv20);
 329
 330             /* Compute parameters for interactions between i and j atoms */
 331             qq20             = _mm_mul_pd(iq2,jq0);
 332
 333             /* Calculate table index by multiplying r with table scale and truncate to integer */
 334             rt               = _mm_mul_pd(r20,vftabscale);
 335             vfitab           = _mm_cvttpd_epi32(rt);
 336 #ifdef __XOP__
 337             vfeps            = _mm_frcz_pd(rt);
 338 #else
 339             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 340 #endif
 341             twovfeps         = _mm_add_pd(vfeps,vfeps);
 342             vfitab           = _mm_slli_epi32(vfitab,2);
 343
 344             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 345             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 346             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 347             GMX_MM_TRANSPOSE2_PD(Y,F);
 348             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 349             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 350             GMX_MM_TRANSPOSE2_PD(G,H);
 351             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 352             VV               = _mm_macc_pd(vfeps,Fp,Y);
 353             velec            = _mm_mul_pd(qq20,VV);
 354             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 355             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq20,FF),_mm_mul_pd(vftabscale,rinv20)));
 356
 357             /* Update potential sum for this i atom from the interaction with this j atom. */
 358             velecsum         = _mm_add_pd(velecsum,velec);
 359
 360             fscal            = felec;
 361
 362             /* Update vectorial force */
 363             fix2             = _mm_macc_pd(dx20,fscal,fix2);
 364             fiy2             = _mm_macc_pd(dy20,fscal,fiy2);
 365             fiz2             = _mm_macc_pd(dz20,fscal,fiz2);
 366
 367             fjx0             = _mm_macc_pd(dx20,fscal,fjx0);
 368             fjy0             = _mm_macc_pd(dy20,fscal,fjy0);
 369             fjz0             = _mm_macc_pd(dz20,fscal,fjz0);
 370
 371             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
 372
 373             /* Inner loop uses 154 flops */
 374         }
 375
 376         if(jidx<j_index_end)
 377         {
 378
 379             jnrA             = jjnr[jidx];
 380             j_coord_offsetA  = DIM*jnrA;
 381
 382             /* load j atom coordinates */
 383             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 384                                               &jx0,&jy0,&jz0);
 385
 386             /* Calculate displacement vector */
 387             dx00             = _mm_sub_pd(ix0,jx0);
 388             dy00             = _mm_sub_pd(iy0,jy0);
 389             dz00             = _mm_sub_pd(iz0,jz0);
 390             dx10             = _mm_sub_pd(ix1,jx0);
 391             dy10             = _mm_sub_pd(iy1,jy0);
 392             dz10             = _mm_sub_pd(iz1,jz0);
 393             dx20             = _mm_sub_pd(ix2,jx0);
 394             dy20             = _mm_sub_pd(iy2,jy0);
 395             dz20             = _mm_sub_pd(iz2,jz0);
 396
 397             /* Calculate squared distance and things based on it */
 398             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 399             rsq10            = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
 400             rsq20            = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
 401
 402             rinv00           = gmx_mm_invsqrt_pd(rsq00);
 403             rinv10           = gmx_mm_invsqrt_pd(rsq10);
 404             rinv20           = gmx_mm_invsqrt_pd(rsq20);
 405
 406             rinvsq00         = _mm_mul_pd(rinv00,rinv00);
 407
 408             /* Load parameters for j particles */
 409             jq0              = _mm_load_sd(charge+jnrA+0);
 410             vdwjidx0A        = 2*vdwtype[jnrA+0];
 411
 412             fjx0             = _mm_setzero_pd();
 413             fjy0             = _mm_setzero_pd();
 414             fjz0             = _mm_setzero_pd();
 415
 416             /**************************
 417              * CALCULATE INTERACTIONS *
 418              **************************/
 419
 420             r00              = _mm_mul_pd(rsq00,rinv00);
 421
 422             /* Compute parameters for interactions between i and j atoms */
 423             qq00             = _mm_mul_pd(iq0,jq0);
 424             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 425
 426             /* Calculate table index by multiplying r with table scale and truncate to integer */
 427             rt               = _mm_mul_pd(r00,vftabscale);
 428             vfitab           = _mm_cvttpd_epi32(rt);
 429 #ifdef __XOP__
 430             vfeps            = _mm_frcz_pd(rt);
 431 #else
 432             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 433 #endif
 434             twovfeps         = _mm_add_pd(vfeps,vfeps);
 435             vfitab           = _mm_slli_epi32(vfitab,2);
 436
 437             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 438             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 439             F                = _mm_setzero_pd();
 440             GMX_MM_TRANSPOSE2_PD(Y,F);
 441             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 442             H                = _mm_setzero_pd();
 443             GMX_MM_TRANSPOSE2_PD(G,H);
 444             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 445             VV               = _mm_macc_pd(vfeps,Fp,Y);
 446             velec            = _mm_mul_pd(qq00,VV);
 447             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 448             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq00,FF),_mm_mul_pd(vftabscale,rinv00)));
 449
 450             /* LENNARD-JONES DISPERSION/REPULSION */
 451
 452             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 453             vvdw6            = _mm_mul_pd(c6_00,rinvsix);
 454             vvdw12           = _mm_mul_pd(c12_00,_mm_mul_pd(rinvsix,rinvsix));
 455             vvdw             = _mm_msub_pd( vvdw12,one_twelfth, _mm_mul_pd(vvdw6,one_sixth) );
 456             fvdw             = _mm_mul_pd(_mm_sub_pd(vvdw12,vvdw6),rinvsq00);
 457
 458             /* Update potential sum for this i atom from the interaction with this j atom. */
 459             velec            = _mm_unpacklo_pd(velec,_mm_setzero_pd());
 460             velecsum         = _mm_add_pd(velecsum,velec);
 461             vvdw             = _mm_unpacklo_pd(vvdw,_mm_setzero_pd());
 462             vvdwsum          = _mm_add_pd(vvdwsum,vvdw);
 463
 464             fscal            = _mm_add_pd(felec,fvdw);
 465
 466             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 467
 468             /* Update vectorial force */
 469             fix0             = _mm_macc_pd(dx00,fscal,fix0);
 470             fiy0             = _mm_macc_pd(dy00,fscal,fiy0);
 471             fiz0             = _mm_macc_pd(dz00,fscal,fiz0);
 472
 473             fjx0             = _mm_macc_pd(dx00,fscal,fjx0);
 474             fjy0             = _mm_macc_pd(dy00,fscal,fjy0);
 475             fjz0             = _mm_macc_pd(dz00,fscal,fjz0);
 476
 477             /**************************
 478              * CALCULATE INTERACTIONS *
 479              **************************/
 480
 481             r10              = _mm_mul_pd(rsq10,rinv10);
 482
 483             /* Compute parameters for interactions between i and j atoms */
 484             qq10             = _mm_mul_pd(iq1,jq0);
 485
 486             /* Calculate table index by multiplying r with table scale and truncate to integer */
 487             rt               = _mm_mul_pd(r10,vftabscale);
 488             vfitab           = _mm_cvttpd_epi32(rt);
 489 #ifdef __XOP__
 490             vfeps            = _mm_frcz_pd(rt);
 491 #else
 492             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 493 #endif
 494             twovfeps         = _mm_add_pd(vfeps,vfeps);
 495             vfitab           = _mm_slli_epi32(vfitab,2);
 496
 497             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 498             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 499             F                = _mm_setzero_pd();
 500             GMX_MM_TRANSPOSE2_PD(Y,F);
 501             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 502             H                = _mm_setzero_pd();
 503             GMX_MM_TRANSPOSE2_PD(G,H);
 504             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 505             VV               = _mm_macc_pd(vfeps,Fp,Y);
 506             velec            = _mm_mul_pd(qq10,VV);
 507             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 508             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq10,FF),_mm_mul_pd(vftabscale,rinv10)));
 509
 510             /* Update potential sum for this i atom from the interaction with this j atom. */
 511             velec            = _mm_unpacklo_pd(velec,_mm_setzero_pd());
 512             velecsum         = _mm_add_pd(velecsum,velec);
 513
 514             fscal            = felec;
 515
 516             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 517
 518             /* Update vectorial force */
 519             fix1             = _mm_macc_pd(dx10,fscal,fix1);
 520             fiy1             = _mm_macc_pd(dy10,fscal,fiy1);
 521             fiz1             = _mm_macc_pd(dz10,fscal,fiz1);
 522
 523             fjx0             = _mm_macc_pd(dx10,fscal,fjx0);
 524             fjy0             = _mm_macc_pd(dy10,fscal,fjy0);
 525             fjz0             = _mm_macc_pd(dz10,fscal,fjz0);
 526
 527             /**************************
 528              * CALCULATE INTERACTIONS *
 529              **************************/
 530
 531             r20              = _mm_mul_pd(rsq20,rinv20);
 532
 533             /* Compute parameters for interactions between i and j atoms */
 534             qq20             = _mm_mul_pd(iq2,jq0);
 535
 536             /* Calculate table index by multiplying r with table scale and truncate to integer */
 537             rt               = _mm_mul_pd(r20,vftabscale);
 538             vfitab           = _mm_cvttpd_epi32(rt);
 539 #ifdef __XOP__
 540             vfeps            = _mm_frcz_pd(rt);
 541 #else
 542             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 543 #endif
 544             twovfeps         = _mm_add_pd(vfeps,vfeps);
 545             vfitab           = _mm_slli_epi32(vfitab,2);
 546
 547             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 548             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 549             F                = _mm_setzero_pd();
 550             GMX_MM_TRANSPOSE2_PD(Y,F);
 551             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 552             H                = _mm_setzero_pd();
 553             GMX_MM_TRANSPOSE2_PD(G,H);
 554             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 555             VV               = _mm_macc_pd(vfeps,Fp,Y);
 556             velec            = _mm_mul_pd(qq20,VV);
 557             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 558             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq20,FF),_mm_mul_pd(vftabscale,rinv20)));
 559
 560             /* Update potential sum for this i atom from the interaction with this j atom. */
 561             velec            = _mm_unpacklo_pd(velec,_mm_setzero_pd());
 562             velecsum         = _mm_add_pd(velecsum,velec);
 563
 564             fscal            = felec;
 565
 566             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 567
 568             /* Update vectorial force */
 569             fix2             = _mm_macc_pd(dx20,fscal,fix2);
 570             fiy2             = _mm_macc_pd(dy20,fscal,fiy2);
 571             fiz2             = _mm_macc_pd(dz20,fscal,fiz2);
 572
 573             fjx0             = _mm_macc_pd(dx20,fscal,fjx0);
 574             fjy0             = _mm_macc_pd(dy20,fscal,fjy0);
 575             fjz0             = _mm_macc_pd(dz20,fscal,fjz0);
 576
 577             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
 578
 579             /* Inner loop uses 154 flops */
 580         }
 581
 582         /* End of innermost loop */
 583
 584         gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 585                                               f+i_coord_offset,fshift+i_shift_offset);
 586
 587         ggid                        = gid[iidx];
 588         /* Update potential energies */
 589         gmx_mm_update_1pot_pd(velecsum,kernel_data->energygrp_elec+ggid);
 590         gmx_mm_update_1pot_pd(vvdwsum,kernel_data->energygrp_vdw+ggid);
 591
 592         /* Increment number of inner iterations */
 593         inneriter                  += j_index_end - j_index_start;
 594
 595         /* Outer loop uses 20 flops */
 596     }
 597
 598     /* Increment number of outer iterations */
 599     outeriter        += nri;
 600
 601     /* Update outer/inner flops */
 602
 603     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_VF,outeriter*20 + inneriter*154);
 604 }
 605 /*
 606  * Gromacs nonbonded kernel:   nb_kernel_ElecCSTab_VdwLJ_GeomW3P1_F_avx_128_fma_double
 607  * Electrostatics interaction: CubicSplineTable
 608  * VdW interaction:            LennardJones
 609  * Geometry:                   Water3-Particle
 610  * Calculate force/pot:        Force
 611  */
 612 void
 613 nb_kernel_ElecCSTab_VdwLJ_GeomW3P1_F_avx_128_fma_double
 614                     (t_nblist                    * gmx_restrict       nlist,
 615                      rvec                        * gmx_restrict          xx,
 616                      rvec                        * gmx_restrict          ff,
 617                      t_forcerec                  * gmx_restrict          fr,
 618                      t_mdatoms                   * gmx_restrict     mdatoms,
 619                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 620                      t_nrnb                      * gmx_restrict        nrnb)
 621 {
 622     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 623      * just 0 for non-waters.
 624      * Suffixes A,B refer to j loop unrolling done with SSE double precision, e.g. for the two different
 625      * jnr indices corresponding to data put in the four positions in the SIMD register.
 626      */
 627     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 628     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 629     int              jnrA,jnrB;
 630     int              j_coord_offsetA,j_coord_offsetB;
 631     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 632     real             rcutoff_scalar;
 633     real             *shiftvec,*fshift,*x,*f;
 634     __m128d          tx,ty,tz,fscal,rcutoff,rcutoff2,jidxall;
 635     int              vdwioffset0;
 636     __m128d          ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 637     int              vdwioffset1;
 638     __m128d          ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 639     int              vdwioffset2;
 640     __m128d          ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 641     int              vdwjidx0A,vdwjidx0B;
 642     __m128d          jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 643     __m128d          dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 644     __m128d          dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 645     __m128d          dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 646     __m128d          velec,felec,velecsum,facel,crf,krf,krf2;
 647     real             *charge;
 648     int              nvdwtype;
 649     __m128d          rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,sh_vdw_invrcut6;
 650     int              *vdwtype;
 651     real             *vdwparam;
 652     __m128d          one_sixth   = _mm_set1_pd(1.0/6.0);
 653     __m128d          one_twelfth = _mm_set1_pd(1.0/12.0);
 654     __m128i          vfitab;
 655     __m128i          ifour       = _mm_set1_epi32(4);
 656     __m128d          rt,vfeps,vftabscale,Y,F,G,H,Heps,Fp,VV,FF,twovfeps;
 657     real             *vftab;
 658     __m128d          dummy_mask,cutoff_mask;
 659     __m128d          signbit   = gmx_mm_castsi128_pd( _mm_set_epi32(0x80000000,0x00000000,0x80000000,0x00000000) );
 660     __m128d          one     = _mm_set1_pd(1.0);
 661     __m128d          two     = _mm_set1_pd(2.0);
 662     x                = xx[0];
 663     f                = ff[0];
 664
 665     nri              = nlist->nri;
 666     iinr             = nlist->iinr;
 667     jindex           = nlist->jindex;
 668     jjnr             = nlist->jjnr;
 669     shiftidx         = nlist->shift;
 670     gid              = nlist->gid;
 671     shiftvec         = fr->shift_vec[0];
 672     fshift           = fr->fshift[0];
 673     facel            = _mm_set1_pd(fr->epsfac);
 674     charge           = mdatoms->chargeA;
 675     nvdwtype         = fr->ntype;
 676     vdwparam         = fr->nbfp;
 677     vdwtype          = mdatoms->typeA;
 678
 679     vftab            = kernel_data->table_elec->data;
 680     vftabscale       = _mm_set1_pd(kernel_data->table_elec->scale);
 681
 682     /* Setup water-specific parameters */
 683     inr              = nlist->iinr[0];
 684     iq0              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+0]));
 685     iq1              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+1]));
 686     iq2              = _mm_mul_pd(facel,_mm_set1_pd(charge[inr+2]));
 687     vdwioffset0      = 2*nvdwtype*vdwtype[inr+0];
 688
 689     /* Avoid stupid compiler warnings */
 690     jnrA = jnrB = 0;
 691     j_coord_offsetA = 0;
 692     j_coord_offsetB = 0;
 693
 694     outeriter        = 0;
 695     inneriter        = 0;
 696
 697     /* Start outer loop over neighborlists */
 698     for(iidx=0; iidx<nri; iidx++)
 699     {
 700         /* Load shift vector for this list */
 701         i_shift_offset   = DIM*shiftidx[iidx];
 702
 703         /* Load limits for loop over neighbors */
 704         j_index_start    = jindex[iidx];
 705         j_index_end      = jindex[iidx+1];
 706
 707         /* Get outer coordinate index */
 708         inr              = iinr[iidx];
 709         i_coord_offset   = DIM*inr;
 710
 711         /* Load i particle coords and add shift vector */
 712         gmx_mm_load_shift_and_3rvec_broadcast_pd(shiftvec+i_shift_offset,x+i_coord_offset,
 713                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 714
 715         fix0             = _mm_setzero_pd();
 716         fiy0             = _mm_setzero_pd();
 717         fiz0             = _mm_setzero_pd();
 718         fix1             = _mm_setzero_pd();
 719         fiy1             = _mm_setzero_pd();
 720         fiz1             = _mm_setzero_pd();
 721         fix2             = _mm_setzero_pd();
 722         fiy2             = _mm_setzero_pd();
 723         fiz2             = _mm_setzero_pd();
 724
 725         /* Start inner kernel loop */
 726         for(jidx=j_index_start; jidx<j_index_end-1; jidx+=2)
 727         {
 728
 729             /* Get j neighbor index, and coordinate index */
 730             jnrA             = jjnr[jidx];
 731             jnrB             = jjnr[jidx+1];
 732             j_coord_offsetA  = DIM*jnrA;
 733             j_coord_offsetB  = DIM*jnrB;
 734
 735             /* load j atom coordinates */
 736             gmx_mm_load_1rvec_2ptr_swizzle_pd(x+j_coord_offsetA,x+j_coord_offsetB,
 737                                               &jx0,&jy0,&jz0);
 738
 739             /* Calculate displacement vector */
 740             dx00             = _mm_sub_pd(ix0,jx0);
 741             dy00             = _mm_sub_pd(iy0,jy0);
 742             dz00             = _mm_sub_pd(iz0,jz0);
 743             dx10             = _mm_sub_pd(ix1,jx0);
 744             dy10             = _mm_sub_pd(iy1,jy0);
 745             dz10             = _mm_sub_pd(iz1,jz0);
 746             dx20             = _mm_sub_pd(ix2,jx0);
 747             dy20             = _mm_sub_pd(iy2,jy0);
 748             dz20             = _mm_sub_pd(iz2,jz0);
 749
 750             /* Calculate squared distance and things based on it */
 751             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 752             rsq10            = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
 753             rsq20            = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
 754
 755             rinv00           = gmx_mm_invsqrt_pd(rsq00);
 756             rinv10           = gmx_mm_invsqrt_pd(rsq10);
 757             rinv20           = gmx_mm_invsqrt_pd(rsq20);
 758
 759             rinvsq00         = _mm_mul_pd(rinv00,rinv00);
 760
 761             /* Load parameters for j particles */
 762             jq0              = gmx_mm_load_2real_swizzle_pd(charge+jnrA+0,charge+jnrB+0);
 763             vdwjidx0A        = 2*vdwtype[jnrA+0];
 764             vdwjidx0B        = 2*vdwtype[jnrB+0];
 765
 766             fjx0             = _mm_setzero_pd();
 767             fjy0             = _mm_setzero_pd();
 768             fjz0             = _mm_setzero_pd();
 769
 770             /**************************
 771              * CALCULATE INTERACTIONS *
 772              **************************/
 773
 774             r00              = _mm_mul_pd(rsq00,rinv00);
 775
 776             /* Compute parameters for interactions between i and j atoms */
 777             qq00             = _mm_mul_pd(iq0,jq0);
 778             gmx_mm_load_2pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,
 779                                          vdwparam+vdwioffset0+vdwjidx0B,&c6_00,&c12_00);
 780
 781             /* Calculate table index by multiplying r with table scale and truncate to integer */
 782             rt               = _mm_mul_pd(r00,vftabscale);
 783             vfitab           = _mm_cvttpd_epi32(rt);
 784 #ifdef __XOP__
 785             vfeps            = _mm_frcz_pd(rt);
 786 #else
 787             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 788 #endif
 789             twovfeps         = _mm_add_pd(vfeps,vfeps);
 790             vfitab           = _mm_slli_epi32(vfitab,2);
 791
 792             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 793             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 794             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 795             GMX_MM_TRANSPOSE2_PD(Y,F);
 796             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 797             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 798             GMX_MM_TRANSPOSE2_PD(G,H);
 799             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 800             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 801             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq00,FF),_mm_mul_pd(vftabscale,rinv00)));
 802
 803             /* LENNARD-JONES DISPERSION/REPULSION */
 804
 805             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 806             fvdw             = _mm_mul_pd(_mm_msub_pd(c12_00,rinvsix,c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 807
 808             fscal            = _mm_add_pd(felec,fvdw);
 809
 810             /* Update vectorial force */
 811             fix0             = _mm_macc_pd(dx00,fscal,fix0);
 812             fiy0             = _mm_macc_pd(dy00,fscal,fiy0);
 813             fiz0             = _mm_macc_pd(dz00,fscal,fiz0);
 814
 815             fjx0             = _mm_macc_pd(dx00,fscal,fjx0);
 816             fjy0             = _mm_macc_pd(dy00,fscal,fjy0);
 817             fjz0             = _mm_macc_pd(dz00,fscal,fjz0);
 818
 819             /**************************
 820              * CALCULATE INTERACTIONS *
 821              **************************/
 822
 823             r10              = _mm_mul_pd(rsq10,rinv10);
 824
 825             /* Compute parameters for interactions between i and j atoms */
 826             qq10             = _mm_mul_pd(iq1,jq0);
 827
 828             /* Calculate table index by multiplying r with table scale and truncate to integer */
 829             rt               = _mm_mul_pd(r10,vftabscale);
 830             vfitab           = _mm_cvttpd_epi32(rt);
 831 #ifdef __XOP__
 832             vfeps            = _mm_frcz_pd(rt);
 833 #else
 834             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 835 #endif
 836             twovfeps         = _mm_add_pd(vfeps,vfeps);
 837             vfitab           = _mm_slli_epi32(vfitab,2);
 838
 839             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 840             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 841             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 842             GMX_MM_TRANSPOSE2_PD(Y,F);
 843             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 844             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 845             GMX_MM_TRANSPOSE2_PD(G,H);
 846             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 847             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 848             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq10,FF),_mm_mul_pd(vftabscale,rinv10)));
 849
 850             fscal            = felec;
 851
 852             /* Update vectorial force */
 853             fix1             = _mm_macc_pd(dx10,fscal,fix1);
 854             fiy1             = _mm_macc_pd(dy10,fscal,fiy1);
 855             fiz1             = _mm_macc_pd(dz10,fscal,fiz1);
 856
 857             fjx0             = _mm_macc_pd(dx10,fscal,fjx0);
 858             fjy0             = _mm_macc_pd(dy10,fscal,fjy0);
 859             fjz0             = _mm_macc_pd(dz10,fscal,fjz0);
 860
 861             /**************************
 862              * CALCULATE INTERACTIONS *
 863              **************************/
 864
 865             r20              = _mm_mul_pd(rsq20,rinv20);
 866
 867             /* Compute parameters for interactions between i and j atoms */
 868             qq20             = _mm_mul_pd(iq2,jq0);
 869
 870             /* Calculate table index by multiplying r with table scale and truncate to integer */
 871             rt               = _mm_mul_pd(r20,vftabscale);
 872             vfitab           = _mm_cvttpd_epi32(rt);
 873 #ifdef __XOP__
 874             vfeps            = _mm_frcz_pd(rt);
 875 #else
 876             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 877 #endif
 878             twovfeps         = _mm_add_pd(vfeps,vfeps);
 879             vfitab           = _mm_slli_epi32(vfitab,2);
 880
 881             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 882             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 883             F                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) );
 884             GMX_MM_TRANSPOSE2_PD(Y,F);
 885             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 886             H                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,1) +2);
 887             GMX_MM_TRANSPOSE2_PD(G,H);
 888             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 889             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 890             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq20,FF),_mm_mul_pd(vftabscale,rinv20)));
 891
 892             fscal            = felec;
 893
 894             /* Update vectorial force */
 895             fix2             = _mm_macc_pd(dx20,fscal,fix2);
 896             fiy2             = _mm_macc_pd(dy20,fscal,fiy2);
 897             fiz2             = _mm_macc_pd(dz20,fscal,fiz2);
 898
 899             fjx0             = _mm_macc_pd(dx20,fscal,fjx0);
 900             fjy0             = _mm_macc_pd(dy20,fscal,fjy0);
 901             fjz0             = _mm_macc_pd(dz20,fscal,fjz0);
 902
 903             gmx_mm_decrement_1rvec_2ptr_swizzle_pd(f+j_coord_offsetA,f+j_coord_offsetB,fjx0,fjy0,fjz0);
 904
 905             /* Inner loop uses 137 flops */
 906         }
 907
 908         if(jidx<j_index_end)
 909         {
 910
 911             jnrA             = jjnr[jidx];
 912             j_coord_offsetA  = DIM*jnrA;
 913
 914             /* load j atom coordinates */
 915             gmx_mm_load_1rvec_1ptr_swizzle_pd(x+j_coord_offsetA,
 916                                               &jx0,&jy0,&jz0);
 917
 918             /* Calculate displacement vector */
 919             dx00             = _mm_sub_pd(ix0,jx0);
 920             dy00             = _mm_sub_pd(iy0,jy0);
 921             dz00             = _mm_sub_pd(iz0,jz0);
 922             dx10             = _mm_sub_pd(ix1,jx0);
 923             dy10             = _mm_sub_pd(iy1,jy0);
 924             dz10             = _mm_sub_pd(iz1,jz0);
 925             dx20             = _mm_sub_pd(ix2,jx0);
 926             dy20             = _mm_sub_pd(iy2,jy0);
 927             dz20             = _mm_sub_pd(iz2,jz0);
 928
 929             /* Calculate squared distance and things based on it */
 930             rsq00            = gmx_mm_calc_rsq_pd(dx00,dy00,dz00);
 931             rsq10            = gmx_mm_calc_rsq_pd(dx10,dy10,dz10);
 932             rsq20            = gmx_mm_calc_rsq_pd(dx20,dy20,dz20);
 933
 934             rinv00           = gmx_mm_invsqrt_pd(rsq00);
 935             rinv10           = gmx_mm_invsqrt_pd(rsq10);
 936             rinv20           = gmx_mm_invsqrt_pd(rsq20);
 937
 938             rinvsq00         = _mm_mul_pd(rinv00,rinv00);
 939
 940             /* Load parameters for j particles */
 941             jq0              = _mm_load_sd(charge+jnrA+0);
 942             vdwjidx0A        = 2*vdwtype[jnrA+0];
 943
 944             fjx0             = _mm_setzero_pd();
 945             fjy0             = _mm_setzero_pd();
 946             fjz0             = _mm_setzero_pd();
 947
 948             /**************************
 949              * CALCULATE INTERACTIONS *
 950              **************************/
 951
 952             r00              = _mm_mul_pd(rsq00,rinv00);
 953
 954             /* Compute parameters for interactions between i and j atoms */
 955             qq00             = _mm_mul_pd(iq0,jq0);
 956             gmx_mm_load_1pair_swizzle_pd(vdwparam+vdwioffset0+vdwjidx0A,&c6_00,&c12_00);
 957
 958             /* Calculate table index by multiplying r with table scale and truncate to integer */
 959             rt               = _mm_mul_pd(r00,vftabscale);
 960             vfitab           = _mm_cvttpd_epi32(rt);
 961 #ifdef __XOP__
 962             vfeps            = _mm_frcz_pd(rt);
 963 #else
 964             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
 965 #endif
 966             twovfeps         = _mm_add_pd(vfeps,vfeps);
 967             vfitab           = _mm_slli_epi32(vfitab,2);
 968
 969             /* CUBIC SPLINE TABLE ELECTROSTATICS */
 970             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
 971             F                = _mm_setzero_pd();
 972             GMX_MM_TRANSPOSE2_PD(Y,F);
 973             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
 974             H                = _mm_setzero_pd();
 975             GMX_MM_TRANSPOSE2_PD(G,H);
 976             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
 977             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
 978             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq00,FF),_mm_mul_pd(vftabscale,rinv00)));
 979
 980             /* LENNARD-JONES DISPERSION/REPULSION */
 981
 982             rinvsix          = _mm_mul_pd(_mm_mul_pd(rinvsq00,rinvsq00),rinvsq00);
 983             fvdw             = _mm_mul_pd(_mm_msub_pd(c12_00,rinvsix,c6_00),_mm_mul_pd(rinvsix,rinvsq00));
 984
 985             fscal            = _mm_add_pd(felec,fvdw);
 986
 987             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
 988
 989             /* Update vectorial force */
 990             fix0             = _mm_macc_pd(dx00,fscal,fix0);
 991             fiy0             = _mm_macc_pd(dy00,fscal,fiy0);
 992             fiz0             = _mm_macc_pd(dz00,fscal,fiz0);
 993
 994             fjx0             = _mm_macc_pd(dx00,fscal,fjx0);
 995             fjy0             = _mm_macc_pd(dy00,fscal,fjy0);
 996             fjz0             = _mm_macc_pd(dz00,fscal,fjz0);
 997
 998             /**************************
 999              * CALCULATE INTERACTIONS *
1000              **************************/
1001
1002             r10              = _mm_mul_pd(rsq10,rinv10);
1003
1004             /* Compute parameters for interactions between i and j atoms */
1005             qq10             = _mm_mul_pd(iq1,jq0);
1006
1007             /* Calculate table index by multiplying r with table scale and truncate to integer */
1008             rt               = _mm_mul_pd(r10,vftabscale);
1009             vfitab           = _mm_cvttpd_epi32(rt);
1010 #ifdef __XOP__
1011             vfeps            = _mm_frcz_pd(rt);
1012 #else
1013             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
1014 #endif
1015             twovfeps         = _mm_add_pd(vfeps,vfeps);
1016             vfitab           = _mm_slli_epi32(vfitab,2);
1017
1018             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1019             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1020             F                = _mm_setzero_pd();
1021             GMX_MM_TRANSPOSE2_PD(Y,F);
1022             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1023             H                = _mm_setzero_pd();
1024             GMX_MM_TRANSPOSE2_PD(G,H);
1025             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
1026             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
1027             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq10,FF),_mm_mul_pd(vftabscale,rinv10)));
1028
1029             fscal            = felec;
1030
1031             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1032
1033             /* Update vectorial force */
1034             fix1             = _mm_macc_pd(dx10,fscal,fix1);
1035             fiy1             = _mm_macc_pd(dy10,fscal,fiy1);
1036             fiz1             = _mm_macc_pd(dz10,fscal,fiz1);
1037
1038             fjx0             = _mm_macc_pd(dx10,fscal,fjx0);
1039             fjy0             = _mm_macc_pd(dy10,fscal,fjy0);
1040             fjz0             = _mm_macc_pd(dz10,fscal,fjz0);
1041
1042             /**************************
1043              * CALCULATE INTERACTIONS *
1044              **************************/
1045
1046             r20              = _mm_mul_pd(rsq20,rinv20);
1047
1048             /* Compute parameters for interactions between i and j atoms */
1049             qq20             = _mm_mul_pd(iq2,jq0);
1050
1051             /* Calculate table index by multiplying r with table scale and truncate to integer */
1052             rt               = _mm_mul_pd(r20,vftabscale);
1053             vfitab           = _mm_cvttpd_epi32(rt);
1054 #ifdef __XOP__
1055             vfeps            = _mm_frcz_pd(rt);
1056 #else
1057             vfeps            = _mm_sub_pd(rt,_mm_round_pd(rt, _MM_FROUND_FLOOR));
1058 #endif
1059             twovfeps         = _mm_add_pd(vfeps,vfeps);
1060             vfitab           = _mm_slli_epi32(vfitab,2);
1061
1062             /* CUBIC SPLINE TABLE ELECTROSTATICS */
1063             Y                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) );
1064             F                = _mm_setzero_pd();
1065             GMX_MM_TRANSPOSE2_PD(Y,F);
1066             G                = _mm_load_pd( vftab + _mm_extract_epi32(vfitab,0) +2);
1067             H                = _mm_setzero_pd();
1068             GMX_MM_TRANSPOSE2_PD(G,H);
1069             Fp               = _mm_macc_pd(vfeps,_mm_macc_pd(vfeps,H,G),F);
1070             FF               = _mm_macc_pd(_mm_macc_pd(twovfeps,H,G),vfeps,Fp);
1071             felec            = _mm_xor_pd(signbit,_mm_mul_pd(_mm_mul_pd(qq20,FF),_mm_mul_pd(vftabscale,rinv20)));
1072
1073             fscal            = felec;
1074
1075             fscal            = _mm_unpacklo_pd(fscal,_mm_setzero_pd());
1076
1077             /* Update vectorial force */
1078             fix2             = _mm_macc_pd(dx20,fscal,fix2);
1079             fiy2             = _mm_macc_pd(dy20,fscal,fiy2);
1080             fiz2             = _mm_macc_pd(dz20,fscal,fiz2);
1081
1082             fjx0             = _mm_macc_pd(dx20,fscal,fjx0);
1083             fjy0             = _mm_macc_pd(dy20,fscal,fjy0);
1084             fjz0             = _mm_macc_pd(dz20,fscal,fjz0);
1085
1086             gmx_mm_decrement_1rvec_1ptr_swizzle_pd(f+j_coord_offsetA,fjx0,fjy0,fjz0);
1087
1088             /* Inner loop uses 137 flops */
1089         }
1090
1091         /* End of innermost loop */
1092
1093         gmx_mm_update_iforce_3atom_swizzle_pd(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1094                                               f+i_coord_offset,fshift+i_shift_offset);
1095
1096         /* Increment number of inner iterations */
1097         inneriter                  += j_index_end - j_index_start;
1098
1099         /* Outer loop uses 18 flops */
1100     }
1101
1102     /* Increment number of outer iterations */
1103     outeriter        += nri;
1104
1105     /* Update outer/inner flops */
1106
1107     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W3_F,outeriter*18 + inneriter*137);
1108 }