src/gromacs/gmxlib/nonbonded/nb_kernel_avx_128_fma_single/nb_kernel_ElecEwSw_VdwNone_GeomW3P1_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 #include "gmxpre.h"
  39
  40 #include "config.h"
  41
  42 #include <math.h>
  43
  44 #include "../nb_kernel.h"
  45 #include "gromacs/legacyheaders/types/simple.h"
  46 #include "gromacs/math/vec.h"
  47 #include "gromacs/legacyheaders/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_ElecEwSw_VdwNone_GeomW3P1_VF_avx_128_fma_single
  54  * Electrostatics interaction: Ewald
  55  * VdW interaction:            None
  56  * Geometry:                   Water3-Particle
  57  * Calculate force/pot:        PotentialAndForce
  58  */
  59 void
  60 nb_kernel_ElecEwSw_VdwNone_GeomW3P1_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              vdwioffset1;
  88     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  89     int              vdwioffset2;
  90     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  91     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  92     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  93     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
  94     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  95     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  96     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
  97     real             *charge;
  98     __m128i          ewitab;
  99     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 100     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 101     real             *ewtab;
 102     __m128           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 103     real             rswitch_scalar,d_scalar;
 104     __m128           dummy_mask,cutoff_mask;
 105     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 106     __m128           one     = _mm_set1_ps(1.0);
 107     __m128           two     = _mm_set1_ps(2.0);
 108     x                = xx[0];
 109     f                = ff[0];
 110
 111     nri              = nlist->nri;
 112     iinr             = nlist->iinr;
 113     jindex           = nlist->jindex;
 114     jjnr             = nlist->jjnr;
 115     shiftidx         = nlist->shift;
 116     gid              = nlist->gid;
 117     shiftvec         = fr->shift_vec[0];
 118     fshift           = fr->fshift[0];
 119     facel            = _mm_set1_ps(fr->epsfac);
 120     charge           = mdatoms->chargeA;
 121
 122     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 123     beta             = _mm_set1_ps(fr->ic->ewaldcoeff_q);
 124     beta2            = _mm_mul_ps(beta,beta);
 125     beta3            = _mm_mul_ps(beta,beta2);
 126     ewtab            = fr->ic->tabq_coul_FDV0;
 127     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 128     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 129
 130     /* Setup water-specific parameters */
 131     inr              = nlist->iinr[0];
 132     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
 133     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 134     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 135
 136     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 137     rcutoff_scalar   = fr->rcoulomb;
 138     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 139     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 140
 141     rswitch_scalar   = fr->rcoulomb_switch;
 142     rswitch          = _mm_set1_ps(rswitch_scalar);
 143     /* Setup switch parameters */
 144     d_scalar         = rcutoff_scalar-rswitch_scalar;
 145     d                = _mm_set1_ps(d_scalar);
 146     swV3             = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 147     swV4             = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 148     swV5             = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 149     swF2             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 150     swF3             = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 151     swF4             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 152
 153     /* Avoid stupid compiler warnings */
 154     jnrA = jnrB = jnrC = jnrD = 0;
 155     j_coord_offsetA = 0;
 156     j_coord_offsetB = 0;
 157     j_coord_offsetC = 0;
 158     j_coord_offsetD = 0;
 159
 160     outeriter        = 0;
 161     inneriter        = 0;
 162
 163     for(iidx=0;iidx<4*DIM;iidx++)
 164     {
 165         scratch[iidx] = 0.0;
 166     }
 167
 168     /* Start outer loop over neighborlists */
 169     for(iidx=0; iidx<nri; iidx++)
 170     {
 171         /* Load shift vector for this list */
 172         i_shift_offset   = DIM*shiftidx[iidx];
 173
 174         /* Load limits for loop over neighbors */
 175         j_index_start    = jindex[iidx];
 176         j_index_end      = jindex[iidx+1];
 177
 178         /* Get outer coordinate index */
 179         inr              = iinr[iidx];
 180         i_coord_offset   = DIM*inr;
 181
 182         /* Load i particle coords and add shift vector */
 183         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 184                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 185
 186         fix0             = _mm_setzero_ps();
 187         fiy0             = _mm_setzero_ps();
 188         fiz0             = _mm_setzero_ps();
 189         fix1             = _mm_setzero_ps();
 190         fiy1             = _mm_setzero_ps();
 191         fiz1             = _mm_setzero_ps();
 192         fix2             = _mm_setzero_ps();
 193         fiy2             = _mm_setzero_ps();
 194         fiz2             = _mm_setzero_ps();
 195
 196         /* Reset potential sums */
 197         velecsum         = _mm_setzero_ps();
 198
 199         /* Start inner kernel loop */
 200         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 201         {
 202
 203             /* Get j neighbor index, and coordinate index */
 204             jnrA             = jjnr[jidx];
 205             jnrB             = jjnr[jidx+1];
 206             jnrC             = jjnr[jidx+2];
 207             jnrD             = jjnr[jidx+3];
 208             j_coord_offsetA  = DIM*jnrA;
 209             j_coord_offsetB  = DIM*jnrB;
 210             j_coord_offsetC  = DIM*jnrC;
 211             j_coord_offsetD  = DIM*jnrD;
 212
 213             /* load j atom coordinates */
 214             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 215                                               x+j_coord_offsetC,x+j_coord_offsetD,
 216                                               &jx0,&jy0,&jz0);
 217
 218             /* Calculate displacement vector */
 219             dx00             = _mm_sub_ps(ix0,jx0);
 220             dy00             = _mm_sub_ps(iy0,jy0);
 221             dz00             = _mm_sub_ps(iz0,jz0);
 222             dx10             = _mm_sub_ps(ix1,jx0);
 223             dy10             = _mm_sub_ps(iy1,jy0);
 224             dz10             = _mm_sub_ps(iz1,jz0);
 225             dx20             = _mm_sub_ps(ix2,jx0);
 226             dy20             = _mm_sub_ps(iy2,jy0);
 227             dz20             = _mm_sub_ps(iz2,jz0);
 228
 229             /* Calculate squared distance and things based on it */
 230             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 231             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 232             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 233
 234             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 235             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 236             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 237
 238             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 239             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 240             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 241
 242             /* Load parameters for j particles */
 243             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 244                                                               charge+jnrC+0,charge+jnrD+0);
 245
 246             fjx0             = _mm_setzero_ps();
 247             fjy0             = _mm_setzero_ps();
 248             fjz0             = _mm_setzero_ps();
 249
 250             /**************************
 251              * CALCULATE INTERACTIONS *
 252              **************************/
 253
 254             if (gmx_mm_any_lt(rsq00,rcutoff2))
 255             {
 256
 257             r00              = _mm_mul_ps(rsq00,rinv00);
 258
 259             /* Compute parameters for interactions between i and j atoms */
 260             qq00             = _mm_mul_ps(iq0,jq0);
 261
 262             /* EWALD ELECTROSTATICS */
 263
 264             /* Analytical PME correction */
 265             zeta2            = _mm_mul_ps(beta2,rsq00);
 266             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 267             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 268             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 269             felec            = _mm_mul_ps(qq00,felec);
 270             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 271             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv00);
 272             velec            = _mm_mul_ps(qq00,velec);
 273
 274             d                = _mm_sub_ps(r00,rswitch);
 275             d                = _mm_max_ps(d,_mm_setzero_ps());
 276             d2               = _mm_mul_ps(d,d);
 277             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 278
 279             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 280
 281             /* Evaluate switch function */
 282             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 283             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
 284             velec            = _mm_mul_ps(velec,sw);
 285             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 286
 287             /* Update potential sum for this i atom from the interaction with this j atom. */
 288             velec            = _mm_and_ps(velec,cutoff_mask);
 289             velecsum         = _mm_add_ps(velecsum,velec);
 290
 291             fscal            = felec;
 292
 293             fscal            = _mm_and_ps(fscal,cutoff_mask);
 294
 295              /* Update vectorial force */
 296             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 297             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 298             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 299
 300             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 301             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 302             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 303
 304             }
 305
 306             /**************************
 307              * CALCULATE INTERACTIONS *
 308              **************************/
 309
 310             if (gmx_mm_any_lt(rsq10,rcutoff2))
 311             {
 312
 313             r10              = _mm_mul_ps(rsq10,rinv10);
 314
 315             /* Compute parameters for interactions between i and j atoms */
 316             qq10             = _mm_mul_ps(iq1,jq0);
 317
 318             /* EWALD ELECTROSTATICS */
 319
 320             /* Analytical PME correction */
 321             zeta2            = _mm_mul_ps(beta2,rsq10);
 322             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 323             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 324             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 325             felec            = _mm_mul_ps(qq10,felec);
 326             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 327             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 328             velec            = _mm_mul_ps(qq10,velec);
 329
 330             d                = _mm_sub_ps(r10,rswitch);
 331             d                = _mm_max_ps(d,_mm_setzero_ps());
 332             d2               = _mm_mul_ps(d,d);
 333             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 334
 335             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 336
 337             /* Evaluate switch function */
 338             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 339             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 340             velec            = _mm_mul_ps(velec,sw);
 341             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 342
 343             /* Update potential sum for this i atom from the interaction with this j atom. */
 344             velec            = _mm_and_ps(velec,cutoff_mask);
 345             velecsum         = _mm_add_ps(velecsum,velec);
 346
 347             fscal            = felec;
 348
 349             fscal            = _mm_and_ps(fscal,cutoff_mask);
 350
 351              /* Update vectorial force */
 352             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 353             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 354             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 355
 356             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 357             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 358             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 359
 360             }
 361
 362             /**************************
 363              * CALCULATE INTERACTIONS *
 364              **************************/
 365
 366             if (gmx_mm_any_lt(rsq20,rcutoff2))
 367             {
 368
 369             r20              = _mm_mul_ps(rsq20,rinv20);
 370
 371             /* Compute parameters for interactions between i and j atoms */
 372             qq20             = _mm_mul_ps(iq2,jq0);
 373
 374             /* EWALD ELECTROSTATICS */
 375
 376             /* Analytical PME correction */
 377             zeta2            = _mm_mul_ps(beta2,rsq20);
 378             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
 379             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 380             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 381             felec            = _mm_mul_ps(qq20,felec);
 382             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 383             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
 384             velec            = _mm_mul_ps(qq20,velec);
 385
 386             d                = _mm_sub_ps(r20,rswitch);
 387             d                = _mm_max_ps(d,_mm_setzero_ps());
 388             d2               = _mm_mul_ps(d,d);
 389             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 390
 391             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 392
 393             /* Evaluate switch function */
 394             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 395             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
 396             velec            = _mm_mul_ps(velec,sw);
 397             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 398
 399             /* Update potential sum for this i atom from the interaction with this j atom. */
 400             velec            = _mm_and_ps(velec,cutoff_mask);
 401             velecsum         = _mm_add_ps(velecsum,velec);
 402
 403             fscal            = felec;
 404
 405             fscal            = _mm_and_ps(fscal,cutoff_mask);
 406
 407              /* Update vectorial force */
 408             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 409             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 410             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 411
 412             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 413             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 414             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 415
 416             }
 417
 418             fjptrA             = f+j_coord_offsetA;
 419             fjptrB             = f+j_coord_offsetB;
 420             fjptrC             = f+j_coord_offsetC;
 421             fjptrD             = f+j_coord_offsetD;
 422
 423             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 424
 425             /* Inner loop uses 159 flops */
 426         }
 427
 428         if(jidx<j_index_end)
 429         {
 430
 431             /* Get j neighbor index, and coordinate index */
 432             jnrlistA         = jjnr[jidx];
 433             jnrlistB         = jjnr[jidx+1];
 434             jnrlistC         = jjnr[jidx+2];
 435             jnrlistD         = jjnr[jidx+3];
 436             /* Sign of each element will be negative for non-real atoms.
 437              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 438              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 439              */
 440             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 441             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 442             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 443             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 444             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 445             j_coord_offsetA  = DIM*jnrA;
 446             j_coord_offsetB  = DIM*jnrB;
 447             j_coord_offsetC  = DIM*jnrC;
 448             j_coord_offsetD  = DIM*jnrD;
 449
 450             /* load j atom coordinates */
 451             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 452                                               x+j_coord_offsetC,x+j_coord_offsetD,
 453                                               &jx0,&jy0,&jz0);
 454
 455             /* Calculate displacement vector */
 456             dx00             = _mm_sub_ps(ix0,jx0);
 457             dy00             = _mm_sub_ps(iy0,jy0);
 458             dz00             = _mm_sub_ps(iz0,jz0);
 459             dx10             = _mm_sub_ps(ix1,jx0);
 460             dy10             = _mm_sub_ps(iy1,jy0);
 461             dz10             = _mm_sub_ps(iz1,jz0);
 462             dx20             = _mm_sub_ps(ix2,jx0);
 463             dy20             = _mm_sub_ps(iy2,jy0);
 464             dz20             = _mm_sub_ps(iz2,jz0);
 465
 466             /* Calculate squared distance and things based on it */
 467             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 468             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 469             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 470
 471             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 472             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 473             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 474
 475             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 476             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 477             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 478
 479             /* Load parameters for j particles */
 480             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 481                                                               charge+jnrC+0,charge+jnrD+0);
 482
 483             fjx0             = _mm_setzero_ps();
 484             fjy0             = _mm_setzero_ps();
 485             fjz0             = _mm_setzero_ps();
 486
 487             /**************************
 488              * CALCULATE INTERACTIONS *
 489              **************************/
 490
 491             if (gmx_mm_any_lt(rsq00,rcutoff2))
 492             {
 493
 494             r00              = _mm_mul_ps(rsq00,rinv00);
 495             r00              = _mm_andnot_ps(dummy_mask,r00);
 496
 497             /* Compute parameters for interactions between i and j atoms */
 498             qq00             = _mm_mul_ps(iq0,jq0);
 499
 500             /* EWALD ELECTROSTATICS */
 501
 502             /* Analytical PME correction */
 503             zeta2            = _mm_mul_ps(beta2,rsq00);
 504             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 505             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 506             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 507             felec            = _mm_mul_ps(qq00,felec);
 508             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 509             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv00);
 510             velec            = _mm_mul_ps(qq00,velec);
 511
 512             d                = _mm_sub_ps(r00,rswitch);
 513             d                = _mm_max_ps(d,_mm_setzero_ps());
 514             d2               = _mm_mul_ps(d,d);
 515             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 516
 517             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 518
 519             /* Evaluate switch function */
 520             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 521             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
 522             velec            = _mm_mul_ps(velec,sw);
 523             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 524
 525             /* Update potential sum for this i atom from the interaction with this j atom. */
 526             velec            = _mm_and_ps(velec,cutoff_mask);
 527             velec            = _mm_andnot_ps(dummy_mask,velec);
 528             velecsum         = _mm_add_ps(velecsum,velec);
 529
 530             fscal            = felec;
 531
 532             fscal            = _mm_and_ps(fscal,cutoff_mask);
 533
 534             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 535
 536              /* Update vectorial force */
 537             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 538             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 539             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 540
 541             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 542             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 543             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 544
 545             }
 546
 547             /**************************
 548              * CALCULATE INTERACTIONS *
 549              **************************/
 550
 551             if (gmx_mm_any_lt(rsq10,rcutoff2))
 552             {
 553
 554             r10              = _mm_mul_ps(rsq10,rinv10);
 555             r10              = _mm_andnot_ps(dummy_mask,r10);
 556
 557             /* Compute parameters for interactions between i and j atoms */
 558             qq10             = _mm_mul_ps(iq1,jq0);
 559
 560             /* EWALD ELECTROSTATICS */
 561
 562             /* Analytical PME correction */
 563             zeta2            = _mm_mul_ps(beta2,rsq10);
 564             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 565             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 566             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 567             felec            = _mm_mul_ps(qq10,felec);
 568             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 569             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 570             velec            = _mm_mul_ps(qq10,velec);
 571
 572             d                = _mm_sub_ps(r10,rswitch);
 573             d                = _mm_max_ps(d,_mm_setzero_ps());
 574             d2               = _mm_mul_ps(d,d);
 575             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 576
 577             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 578
 579             /* Evaluate switch function */
 580             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 581             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 582             velec            = _mm_mul_ps(velec,sw);
 583             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 584
 585             /* Update potential sum for this i atom from the interaction with this j atom. */
 586             velec            = _mm_and_ps(velec,cutoff_mask);
 587             velec            = _mm_andnot_ps(dummy_mask,velec);
 588             velecsum         = _mm_add_ps(velecsum,velec);
 589
 590             fscal            = felec;
 591
 592             fscal            = _mm_and_ps(fscal,cutoff_mask);
 593
 594             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 595
 596              /* Update vectorial force */
 597             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 598             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 599             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 600
 601             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 602             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 603             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 604
 605             }
 606
 607             /**************************
 608              * CALCULATE INTERACTIONS *
 609              **************************/
 610
 611             if (gmx_mm_any_lt(rsq20,rcutoff2))
 612             {
 613
 614             r20              = _mm_mul_ps(rsq20,rinv20);
 615             r20              = _mm_andnot_ps(dummy_mask,r20);
 616
 617             /* Compute parameters for interactions between i and j atoms */
 618             qq20             = _mm_mul_ps(iq2,jq0);
 619
 620             /* EWALD ELECTROSTATICS */
 621
 622             /* Analytical PME correction */
 623             zeta2            = _mm_mul_ps(beta2,rsq20);
 624             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
 625             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 626             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 627             felec            = _mm_mul_ps(qq20,felec);
 628             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 629             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
 630             velec            = _mm_mul_ps(qq20,velec);
 631
 632             d                = _mm_sub_ps(r20,rswitch);
 633             d                = _mm_max_ps(d,_mm_setzero_ps());
 634             d2               = _mm_mul_ps(d,d);
 635             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 636
 637             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 638
 639             /* Evaluate switch function */
 640             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 641             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
 642             velec            = _mm_mul_ps(velec,sw);
 643             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 644
 645             /* Update potential sum for this i atom from the interaction with this j atom. */
 646             velec            = _mm_and_ps(velec,cutoff_mask);
 647             velec            = _mm_andnot_ps(dummy_mask,velec);
 648             velecsum         = _mm_add_ps(velecsum,velec);
 649
 650             fscal            = felec;
 651
 652             fscal            = _mm_and_ps(fscal,cutoff_mask);
 653
 654             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 655
 656              /* Update vectorial force */
 657             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 658             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 659             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 660
 661             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 662             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 663             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 664
 665             }
 666
 667             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 668             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 669             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 670             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 671
 672             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 673
 674             /* Inner loop uses 162 flops */
 675         }
 676
 677         /* End of innermost loop */
 678
 679         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
 680                                               f+i_coord_offset,fshift+i_shift_offset);
 681
 682         ggid                        = gid[iidx];
 683         /* Update potential energies */
 684         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 685
 686         /* Increment number of inner iterations */
 687         inneriter                  += j_index_end - j_index_start;
 688
 689         /* Outer loop uses 19 flops */
 690     }
 691
 692     /* Increment number of outer iterations */
 693     outeriter        += nri;
 694
 695     /* Update outer/inner flops */
 696
 697     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_VF,outeriter*19 + inneriter*162);
 698 }
 699 /*
 700  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwNone_GeomW3P1_F_avx_128_fma_single
 701  * Electrostatics interaction: Ewald
 702  * VdW interaction:            None
 703  * Geometry:                   Water3-Particle
 704  * Calculate force/pot:        Force
 705  */
 706 void
 707 nb_kernel_ElecEwSw_VdwNone_GeomW3P1_F_avx_128_fma_single
 708                     (t_nblist                    * gmx_restrict       nlist,
 709                      rvec                        * gmx_restrict          xx,
 710                      rvec                        * gmx_restrict          ff,
 711                      t_forcerec                  * gmx_restrict          fr,
 712                      t_mdatoms                   * gmx_restrict     mdatoms,
 713                      nb_kernel_data_t gmx_unused * gmx_restrict kernel_data,
 714                      t_nrnb                      * gmx_restrict        nrnb)
 715 {
 716     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 717      * just 0 for non-waters.
 718      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
 719      * jnr indices corresponding to data put in the four positions in the SIMD register.
 720      */
 721     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 722     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 723     int              jnrA,jnrB,jnrC,jnrD;
 724     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 725     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 726     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 727     real             rcutoff_scalar;
 728     real             *shiftvec,*fshift,*x,*f;
 729     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 730     real             scratch[4*DIM];
 731     __m128           fscal,rcutoff,rcutoff2,jidxall;
 732     int              vdwioffset0;
 733     __m128           ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 734     int              vdwioffset1;
 735     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 736     int              vdwioffset2;
 737     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 738     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 739     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 740     __m128           dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00;
 741     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 742     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 743     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 744     real             *charge;
 745     __m128i          ewitab;
 746     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 747     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 748     real             *ewtab;
 749     __m128           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 750     real             rswitch_scalar,d_scalar;
 751     __m128           dummy_mask,cutoff_mask;
 752     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 753     __m128           one     = _mm_set1_ps(1.0);
 754     __m128           two     = _mm_set1_ps(2.0);
 755     x                = xx[0];
 756     f                = ff[0];
 757
 758     nri              = nlist->nri;
 759     iinr             = nlist->iinr;
 760     jindex           = nlist->jindex;
 761     jjnr             = nlist->jjnr;
 762     shiftidx         = nlist->shift;
 763     gid              = nlist->gid;
 764     shiftvec         = fr->shift_vec[0];
 765     fshift           = fr->fshift[0];
 766     facel            = _mm_set1_ps(fr->epsfac);
 767     charge           = mdatoms->chargeA;
 768
 769     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 770     beta             = _mm_set1_ps(fr->ic->ewaldcoeff_q);
 771     beta2            = _mm_mul_ps(beta,beta);
 772     beta3            = _mm_mul_ps(beta,beta2);
 773     ewtab            = fr->ic->tabq_coul_FDV0;
 774     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 775     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 776
 777     /* Setup water-specific parameters */
 778     inr              = nlist->iinr[0];
 779     iq0              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+0]));
 780     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 781     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 782
 783     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 784     rcutoff_scalar   = fr->rcoulomb;
 785     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 786     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 787
 788     rswitch_scalar   = fr->rcoulomb_switch;
 789     rswitch          = _mm_set1_ps(rswitch_scalar);
 790     /* Setup switch parameters */
 791     d_scalar         = rcutoff_scalar-rswitch_scalar;
 792     d                = _mm_set1_ps(d_scalar);
 793     swV3             = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 794     swV4             = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 795     swV5             = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 796     swF2             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 797     swF3             = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 798     swF4             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 799
 800     /* Avoid stupid compiler warnings */
 801     jnrA = jnrB = jnrC = jnrD = 0;
 802     j_coord_offsetA = 0;
 803     j_coord_offsetB = 0;
 804     j_coord_offsetC = 0;
 805     j_coord_offsetD = 0;
 806
 807     outeriter        = 0;
 808     inneriter        = 0;
 809
 810     for(iidx=0;iidx<4*DIM;iidx++)
 811     {
 812         scratch[iidx] = 0.0;
 813     }
 814
 815     /* Start outer loop over neighborlists */
 816     for(iidx=0; iidx<nri; iidx++)
 817     {
 818         /* Load shift vector for this list */
 819         i_shift_offset   = DIM*shiftidx[iidx];
 820
 821         /* Load limits for loop over neighbors */
 822         j_index_start    = jindex[iidx];
 823         j_index_end      = jindex[iidx+1];
 824
 825         /* Get outer coordinate index */
 826         inr              = iinr[iidx];
 827         i_coord_offset   = DIM*inr;
 828
 829         /* Load i particle coords and add shift vector */
 830         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset,
 831                                                  &ix0,&iy0,&iz0,&ix1,&iy1,&iz1,&ix2,&iy2,&iz2);
 832
 833         fix0             = _mm_setzero_ps();
 834         fiy0             = _mm_setzero_ps();
 835         fiz0             = _mm_setzero_ps();
 836         fix1             = _mm_setzero_ps();
 837         fiy1             = _mm_setzero_ps();
 838         fiz1             = _mm_setzero_ps();
 839         fix2             = _mm_setzero_ps();
 840         fiy2             = _mm_setzero_ps();
 841         fiz2             = _mm_setzero_ps();
 842
 843         /* Start inner kernel loop */
 844         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 845         {
 846
 847             /* Get j neighbor index, and coordinate index */
 848             jnrA             = jjnr[jidx];
 849             jnrB             = jjnr[jidx+1];
 850             jnrC             = jjnr[jidx+2];
 851             jnrD             = jjnr[jidx+3];
 852             j_coord_offsetA  = DIM*jnrA;
 853             j_coord_offsetB  = DIM*jnrB;
 854             j_coord_offsetC  = DIM*jnrC;
 855             j_coord_offsetD  = DIM*jnrD;
 856
 857             /* load j atom coordinates */
 858             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 859                                               x+j_coord_offsetC,x+j_coord_offsetD,
 860                                               &jx0,&jy0,&jz0);
 861
 862             /* Calculate displacement vector */
 863             dx00             = _mm_sub_ps(ix0,jx0);
 864             dy00             = _mm_sub_ps(iy0,jy0);
 865             dz00             = _mm_sub_ps(iz0,jz0);
 866             dx10             = _mm_sub_ps(ix1,jx0);
 867             dy10             = _mm_sub_ps(iy1,jy0);
 868             dz10             = _mm_sub_ps(iz1,jz0);
 869             dx20             = _mm_sub_ps(ix2,jx0);
 870             dy20             = _mm_sub_ps(iy2,jy0);
 871             dz20             = _mm_sub_ps(iz2,jz0);
 872
 873             /* Calculate squared distance and things based on it */
 874             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
 875             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 876             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 877
 878             rinv00           = gmx_mm_invsqrt_ps(rsq00);
 879             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 880             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 881
 882             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
 883             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 884             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 885
 886             /* Load parameters for j particles */
 887             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 888                                                               charge+jnrC+0,charge+jnrD+0);
 889
 890             fjx0             = _mm_setzero_ps();
 891             fjy0             = _mm_setzero_ps();
 892             fjz0             = _mm_setzero_ps();
 893
 894             /**************************
 895              * CALCULATE INTERACTIONS *
 896              **************************/
 897
 898             if (gmx_mm_any_lt(rsq00,rcutoff2))
 899             {
 900
 901             r00              = _mm_mul_ps(rsq00,rinv00);
 902
 903             /* Compute parameters for interactions between i and j atoms */
 904             qq00             = _mm_mul_ps(iq0,jq0);
 905
 906             /* EWALD ELECTROSTATICS */
 907
 908             /* Analytical PME correction */
 909             zeta2            = _mm_mul_ps(beta2,rsq00);
 910             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
 911             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 912             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 913             felec            = _mm_mul_ps(qq00,felec);
 914             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 915             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv00);
 916             velec            = _mm_mul_ps(qq00,velec);
 917
 918             d                = _mm_sub_ps(r00,rswitch);
 919             d                = _mm_max_ps(d,_mm_setzero_ps());
 920             d2               = _mm_mul_ps(d,d);
 921             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 922
 923             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 924
 925             /* Evaluate switch function */
 926             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 927             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
 928             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
 929
 930             fscal            = felec;
 931
 932             fscal            = _mm_and_ps(fscal,cutoff_mask);
 933
 934              /* Update vectorial force */
 935             fix0             = _mm_macc_ps(dx00,fscal,fix0);
 936             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
 937             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
 938
 939             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
 940             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
 941             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
 942
 943             }
 944
 945             /**************************
 946              * CALCULATE INTERACTIONS *
 947              **************************/
 948
 949             if (gmx_mm_any_lt(rsq10,rcutoff2))
 950             {
 951
 952             r10              = _mm_mul_ps(rsq10,rinv10);
 953
 954             /* Compute parameters for interactions between i and j atoms */
 955             qq10             = _mm_mul_ps(iq1,jq0);
 956
 957             /* EWALD ELECTROSTATICS */
 958
 959             /* Analytical PME correction */
 960             zeta2            = _mm_mul_ps(beta2,rsq10);
 961             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 962             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 963             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 964             felec            = _mm_mul_ps(qq10,felec);
 965             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 966             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 967             velec            = _mm_mul_ps(qq10,velec);
 968
 969             d                = _mm_sub_ps(r10,rswitch);
 970             d                = _mm_max_ps(d,_mm_setzero_ps());
 971             d2               = _mm_mul_ps(d,d);
 972             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 973
 974             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 975
 976             /* Evaluate switch function */
 977             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 978             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 979             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 980
 981             fscal            = felec;
 982
 983             fscal            = _mm_and_ps(fscal,cutoff_mask);
 984
 985              /* Update vectorial force */
 986             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 987             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 988             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 989
 990             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 991             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 992             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 993
 994             }
 995
 996             /**************************
 997              * CALCULATE INTERACTIONS *
 998              **************************/
 999
1000             if (gmx_mm_any_lt(rsq20,rcutoff2))
1001             {
1002
1003             r20              = _mm_mul_ps(rsq20,rinv20);
1004
1005             /* Compute parameters for interactions between i and j atoms */
1006             qq20             = _mm_mul_ps(iq2,jq0);
1007
1008             /* EWALD ELECTROSTATICS */
1009
1010             /* Analytical PME correction */
1011             zeta2            = _mm_mul_ps(beta2,rsq20);
1012             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
1013             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1014             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1015             felec            = _mm_mul_ps(qq20,felec);
1016             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1017             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
1018             velec            = _mm_mul_ps(qq20,velec);
1019
1020             d                = _mm_sub_ps(r20,rswitch);
1021             d                = _mm_max_ps(d,_mm_setzero_ps());
1022             d2               = _mm_mul_ps(d,d);
1023             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1024
1025             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1026
1027             /* Evaluate switch function */
1028             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1029             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1030             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1031
1032             fscal            = felec;
1033
1034             fscal            = _mm_and_ps(fscal,cutoff_mask);
1035
1036              /* Update vectorial force */
1037             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1038             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1039             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1040
1041             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1042             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1043             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1044
1045             }
1046
1047             fjptrA             = f+j_coord_offsetA;
1048             fjptrB             = f+j_coord_offsetB;
1049             fjptrC             = f+j_coord_offsetC;
1050             fjptrD             = f+j_coord_offsetD;
1051
1052             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1053
1054             /* Inner loop uses 150 flops */
1055         }
1056
1057         if(jidx<j_index_end)
1058         {
1059
1060             /* Get j neighbor index, and coordinate index */
1061             jnrlistA         = jjnr[jidx];
1062             jnrlistB         = jjnr[jidx+1];
1063             jnrlistC         = jjnr[jidx+2];
1064             jnrlistD         = jjnr[jidx+3];
1065             /* Sign of each element will be negative for non-real atoms.
1066              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1067              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1068              */
1069             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1070             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1071             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1072             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1073             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1074             j_coord_offsetA  = DIM*jnrA;
1075             j_coord_offsetB  = DIM*jnrB;
1076             j_coord_offsetC  = DIM*jnrC;
1077             j_coord_offsetD  = DIM*jnrD;
1078
1079             /* load j atom coordinates */
1080             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1081                                               x+j_coord_offsetC,x+j_coord_offsetD,
1082                                               &jx0,&jy0,&jz0);
1083
1084             /* Calculate displacement vector */
1085             dx00             = _mm_sub_ps(ix0,jx0);
1086             dy00             = _mm_sub_ps(iy0,jy0);
1087             dz00             = _mm_sub_ps(iz0,jz0);
1088             dx10             = _mm_sub_ps(ix1,jx0);
1089             dy10             = _mm_sub_ps(iy1,jy0);
1090             dz10             = _mm_sub_ps(iz1,jz0);
1091             dx20             = _mm_sub_ps(ix2,jx0);
1092             dy20             = _mm_sub_ps(iy2,jy0);
1093             dz20             = _mm_sub_ps(iz2,jz0);
1094
1095             /* Calculate squared distance and things based on it */
1096             rsq00            = gmx_mm_calc_rsq_ps(dx00,dy00,dz00);
1097             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1098             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1099
1100             rinv00           = gmx_mm_invsqrt_ps(rsq00);
1101             rinv10           = gmx_mm_invsqrt_ps(rsq10);
1102             rinv20           = gmx_mm_invsqrt_ps(rsq20);
1103
1104             rinvsq00         = _mm_mul_ps(rinv00,rinv00);
1105             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1106             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1107
1108             /* Load parameters for j particles */
1109             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1110                                                               charge+jnrC+0,charge+jnrD+0);
1111
1112             fjx0             = _mm_setzero_ps();
1113             fjy0             = _mm_setzero_ps();
1114             fjz0             = _mm_setzero_ps();
1115
1116             /**************************
1117              * CALCULATE INTERACTIONS *
1118              **************************/
1119
1120             if (gmx_mm_any_lt(rsq00,rcutoff2))
1121             {
1122
1123             r00              = _mm_mul_ps(rsq00,rinv00);
1124             r00              = _mm_andnot_ps(dummy_mask,r00);
1125
1126             /* Compute parameters for interactions between i and j atoms */
1127             qq00             = _mm_mul_ps(iq0,jq0);
1128
1129             /* EWALD ELECTROSTATICS */
1130
1131             /* Analytical PME correction */
1132             zeta2            = _mm_mul_ps(beta2,rsq00);
1133             rinv3            = _mm_mul_ps(rinvsq00,rinv00);
1134             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1135             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1136             felec            = _mm_mul_ps(qq00,felec);
1137             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1138             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv00);
1139             velec            = _mm_mul_ps(qq00,velec);
1140
1141             d                = _mm_sub_ps(r00,rswitch);
1142             d                = _mm_max_ps(d,_mm_setzero_ps());
1143             d2               = _mm_mul_ps(d,d);
1144             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1145
1146             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1147
1148             /* Evaluate switch function */
1149             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1150             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv00,_mm_mul_ps(velec,dsw)) );
1151             cutoff_mask      = _mm_cmplt_ps(rsq00,rcutoff2);
1152
1153             fscal            = felec;
1154
1155             fscal            = _mm_and_ps(fscal,cutoff_mask);
1156
1157             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1158
1159              /* Update vectorial force */
1160             fix0             = _mm_macc_ps(dx00,fscal,fix0);
1161             fiy0             = _mm_macc_ps(dy00,fscal,fiy0);
1162             fiz0             = _mm_macc_ps(dz00,fscal,fiz0);
1163
1164             fjx0             = _mm_macc_ps(dx00,fscal,fjx0);
1165             fjy0             = _mm_macc_ps(dy00,fscal,fjy0);
1166             fjz0             = _mm_macc_ps(dz00,fscal,fjz0);
1167
1168             }
1169
1170             /**************************
1171              * CALCULATE INTERACTIONS *
1172              **************************/
1173
1174             if (gmx_mm_any_lt(rsq10,rcutoff2))
1175             {
1176
1177             r10              = _mm_mul_ps(rsq10,rinv10);
1178             r10              = _mm_andnot_ps(dummy_mask,r10);
1179
1180             /* Compute parameters for interactions between i and j atoms */
1181             qq10             = _mm_mul_ps(iq1,jq0);
1182
1183             /* EWALD ELECTROSTATICS */
1184
1185             /* Analytical PME correction */
1186             zeta2            = _mm_mul_ps(beta2,rsq10);
1187             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
1188             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1189             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1190             felec            = _mm_mul_ps(qq10,felec);
1191             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1192             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
1193             velec            = _mm_mul_ps(qq10,velec);
1194
1195             d                = _mm_sub_ps(r10,rswitch);
1196             d                = _mm_max_ps(d,_mm_setzero_ps());
1197             d2               = _mm_mul_ps(d,d);
1198             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1199
1200             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1201
1202             /* Evaluate switch function */
1203             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1204             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1205             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1206
1207             fscal            = felec;
1208
1209             fscal            = _mm_and_ps(fscal,cutoff_mask);
1210
1211             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1212
1213              /* Update vectorial force */
1214             fix1             = _mm_macc_ps(dx10,fscal,fix1);
1215             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
1216             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
1217
1218             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
1219             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
1220             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
1221
1222             }
1223
1224             /**************************
1225              * CALCULATE INTERACTIONS *
1226              **************************/
1227
1228             if (gmx_mm_any_lt(rsq20,rcutoff2))
1229             {
1230
1231             r20              = _mm_mul_ps(rsq20,rinv20);
1232             r20              = _mm_andnot_ps(dummy_mask,r20);
1233
1234             /* Compute parameters for interactions between i and j atoms */
1235             qq20             = _mm_mul_ps(iq2,jq0);
1236
1237             /* EWALD ELECTROSTATICS */
1238
1239             /* Analytical PME correction */
1240             zeta2            = _mm_mul_ps(beta2,rsq20);
1241             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
1242             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1243             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1244             felec            = _mm_mul_ps(qq20,felec);
1245             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1246             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
1247             velec            = _mm_mul_ps(qq20,velec);
1248
1249             d                = _mm_sub_ps(r20,rswitch);
1250             d                = _mm_max_ps(d,_mm_setzero_ps());
1251             d2               = _mm_mul_ps(d,d);
1252             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1253
1254             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1255
1256             /* Evaluate switch function */
1257             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1258             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1259             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1260
1261             fscal            = felec;
1262
1263             fscal            = _mm_and_ps(fscal,cutoff_mask);
1264
1265             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1266
1267              /* Update vectorial force */
1268             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1269             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1270             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1271
1272             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1273             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1274             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1275
1276             }
1277
1278             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1279             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1280             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1281             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1282
1283             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1284
1285             /* Inner loop uses 153 flops */
1286         }
1287
1288         /* End of innermost loop */
1289
1290         gmx_mm_update_iforce_3atom_swizzle_ps(fix0,fiy0,fiz0,fix1,fiy1,fiz1,fix2,fiy2,fiz2,
1291                                               f+i_coord_offset,fshift+i_shift_offset);
1292
1293         /* Increment number of inner iterations */
1294         inneriter                  += j_index_end - j_index_start;
1295
1296         /* Outer loop uses 18 flops */
1297     }
1298
1299     /* Increment number of outer iterations */
1300     outeriter        += nri;
1301
1302     /* Update outer/inner flops */
1303
1304     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W3_F,outeriter*18 + inneriter*153);
1305 }