src/gmxlib/nonbonded/nb_kernel_avx_128_fma_single/nb_kernel_ElecEwSw_VdwNone_GeomW4P1_avx_128_fma_single.c

   1 /*
   2  * Note: this file was generated by the Gromacs avx_128_fma_single kernel generator.
   3  *
   4  *                This source code is part of
   5  *
   6  *                 G   R   O   M   A   C   S
   7  *
   8  * Copyright (c) 2001-2012, The GROMACS Development Team
   9  *
  10  * Gromacs is a library for molecular simulation and trajectory analysis,
  11  * written by Erik Lindahl, David van der Spoel, Berk Hess, and others - for
  12  * a full list of developers and information, check out http://www.gromacs.org
  13  *
  14  * This program is free software; you can redistribute it and/or modify it under
  15  * the terms of the GNU Lesser General Public License as published by the Free
  16  * Software Foundation; either version 2 of the License, or (at your option) any
  17  * later version.
  18  *
  19  * To help fund GROMACS development, we humbly ask that you cite
  20  * the papers people have written on it - you can find them on the website.
  21  */
  22 #ifdef HAVE_CONFIG_H
  23 #include <config.h>
  24 #endif
  25
  26 #include <math.h>
  27
  28 #include "../nb_kernel.h"
  29 #include "types/simple.h"
  30 #include "vec.h"
  31 #include "nrnb.h"
  32
  33 #include "gmx_math_x86_avx_128_fma_single.h"
  34 #include "kernelutil_x86_avx_128_fma_single.h"
  35
  36 /*
  37  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwNone_GeomW4P1_VF_avx_128_fma_single
  38  * Electrostatics interaction: Ewald
  39  * VdW interaction:            None
  40  * Geometry:                   Water4-Particle
  41  * Calculate force/pot:        PotentialAndForce
  42  */
  43 void
  44 nb_kernel_ElecEwSw_VdwNone_GeomW4P1_VF_avx_128_fma_single
  45                     (t_nblist * gmx_restrict                nlist,
  46                      rvec * gmx_restrict                    xx,
  47                      rvec * gmx_restrict                    ff,
  48                      t_forcerec * gmx_restrict              fr,
  49                      t_mdatoms * gmx_restrict               mdatoms,
  50                      nb_kernel_data_t * gmx_restrict        kernel_data,
  51                      t_nrnb * gmx_restrict                  nrnb)
  52 {
  53     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
  54      * just 0 for non-waters.
  55      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
  56      * jnr indices corresponding to data put in the four positions in the SIMD register.
  57      */
  58     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
  59     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
  60     int              jnrA,jnrB,jnrC,jnrD;
  61     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
  62     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
  63     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  64     real             rcutoff_scalar;
  65     real             *shiftvec,*fshift,*x,*f;
  66     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
  67     real             scratch[4*DIM];
  68     __m128           fscal,rcutoff,rcutoff2,jidxall;
  69     int              vdwioffset1;
  70     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  71     int              vdwioffset2;
  72     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  73     int              vdwioffset3;
  74     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  75     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
  76     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  77     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
  78     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
  79     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
  80     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
  81     real             *charge;
  82     __m128i          ewitab;
  83     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
  84     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
  85     real             *ewtab;
  86     __m128           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  87     real             rswitch_scalar,d_scalar;
  88     __m128           dummy_mask,cutoff_mask;
  89     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
  90     __m128           one     = _mm_set1_ps(1.0);
  91     __m128           two     = _mm_set1_ps(2.0);
  92     x                = xx[0];
  93     f                = ff[0];
  94
  95     nri              = nlist->nri;
  96     iinr             = nlist->iinr;
  97     jindex           = nlist->jindex;
  98     jjnr             = nlist->jjnr;
  99     shiftidx         = nlist->shift;
 100     gid              = nlist->gid;
 101     shiftvec         = fr->shift_vec[0];
 102     fshift           = fr->fshift[0];
 103     facel            = _mm_set1_ps(fr->epsfac);
 104     charge           = mdatoms->chargeA;
 105
 106     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 107     beta             = _mm_set1_ps(fr->ic->ewaldcoeff);
 108     beta2            = _mm_mul_ps(beta,beta);
 109     beta3            = _mm_mul_ps(beta,beta2);
 110     ewtab            = fr->ic->tabq_coul_FDV0;
 111     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 112     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 113
 114     /* Setup water-specific parameters */
 115     inr              = nlist->iinr[0];
 116     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 117     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 118     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 119
 120     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 121     rcutoff_scalar   = fr->rcoulomb;
 122     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 123     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 124
 125     rswitch_scalar   = fr->rcoulomb_switch;
 126     rswitch          = _mm_set1_ps(rswitch_scalar);
 127     /* Setup switch parameters */
 128     d_scalar         = rcutoff_scalar-rswitch_scalar;
 129     d                = _mm_set1_ps(d_scalar);
 130     swV3             = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 131     swV4             = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 132     swV5             = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 133     swF2             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 134     swF3             = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 135     swF4             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 136
 137     /* Avoid stupid compiler warnings */
 138     jnrA = jnrB = jnrC = jnrD = 0;
 139     j_coord_offsetA = 0;
 140     j_coord_offsetB = 0;
 141     j_coord_offsetC = 0;
 142     j_coord_offsetD = 0;
 143
 144     outeriter        = 0;
 145     inneriter        = 0;
 146
 147     for(iidx=0;iidx<4*DIM;iidx++)
 148     {
 149         scratch[iidx] = 0.0;
 150     }
 151
 152     /* Start outer loop over neighborlists */
 153     for(iidx=0; iidx<nri; iidx++)
 154     {
 155         /* Load shift vector for this list */
 156         i_shift_offset   = DIM*shiftidx[iidx];
 157
 158         /* Load limits for loop over neighbors */
 159         j_index_start    = jindex[iidx];
 160         j_index_end      = jindex[iidx+1];
 161
 162         /* Get outer coordinate index */
 163         inr              = iinr[iidx];
 164         i_coord_offset   = DIM*inr;
 165
 166         /* Load i particle coords and add shift vector */
 167         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 168                                                  &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 169
 170         fix1             = _mm_setzero_ps();
 171         fiy1             = _mm_setzero_ps();
 172         fiz1             = _mm_setzero_ps();
 173         fix2             = _mm_setzero_ps();
 174         fiy2             = _mm_setzero_ps();
 175         fiz2             = _mm_setzero_ps();
 176         fix3             = _mm_setzero_ps();
 177         fiy3             = _mm_setzero_ps();
 178         fiz3             = _mm_setzero_ps();
 179
 180         /* Reset potential sums */
 181         velecsum         = _mm_setzero_ps();
 182
 183         /* Start inner kernel loop */
 184         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 185         {
 186
 187             /* Get j neighbor index, and coordinate index */
 188             jnrA             = jjnr[jidx];
 189             jnrB             = jjnr[jidx+1];
 190             jnrC             = jjnr[jidx+2];
 191             jnrD             = jjnr[jidx+3];
 192             j_coord_offsetA  = DIM*jnrA;
 193             j_coord_offsetB  = DIM*jnrB;
 194             j_coord_offsetC  = DIM*jnrC;
 195             j_coord_offsetD  = DIM*jnrD;
 196
 197             /* load j atom coordinates */
 198             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 199                                               x+j_coord_offsetC,x+j_coord_offsetD,
 200                                               &jx0,&jy0,&jz0);
 201
 202             /* Calculate displacement vector */
 203             dx10             = _mm_sub_ps(ix1,jx0);
 204             dy10             = _mm_sub_ps(iy1,jy0);
 205             dz10             = _mm_sub_ps(iz1,jz0);
 206             dx20             = _mm_sub_ps(ix2,jx0);
 207             dy20             = _mm_sub_ps(iy2,jy0);
 208             dz20             = _mm_sub_ps(iz2,jz0);
 209             dx30             = _mm_sub_ps(ix3,jx0);
 210             dy30             = _mm_sub_ps(iy3,jy0);
 211             dz30             = _mm_sub_ps(iz3,jz0);
 212
 213             /* Calculate squared distance and things based on it */
 214             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 215             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 216             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 217
 218             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 219             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 220             rinv30           = gmx_mm_invsqrt_ps(rsq30);
 221
 222             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 223             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 224             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 225
 226             /* Load parameters for j particles */
 227             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 228                                                               charge+jnrC+0,charge+jnrD+0);
 229
 230             fjx0             = _mm_setzero_ps();
 231             fjy0             = _mm_setzero_ps();
 232             fjz0             = _mm_setzero_ps();
 233
 234             /**************************
 235              * CALCULATE INTERACTIONS *
 236              **************************/
 237
 238             if (gmx_mm_any_lt(rsq10,rcutoff2))
 239             {
 240
 241             r10              = _mm_mul_ps(rsq10,rinv10);
 242
 243             /* Compute parameters for interactions between i and j atoms */
 244             qq10             = _mm_mul_ps(iq1,jq0);
 245
 246             /* EWALD ELECTROSTATICS */
 247
 248             /* Analytical PME correction */
 249             zeta2            = _mm_mul_ps(beta2,rsq10);
 250             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 251             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 252             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 253             felec            = _mm_mul_ps(qq10,felec);
 254             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 255             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 256             velec            = _mm_mul_ps(qq10,velec);
 257
 258             d                = _mm_sub_ps(r10,rswitch);
 259             d                = _mm_max_ps(d,_mm_setzero_ps());
 260             d2               = _mm_mul_ps(d,d);
 261             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 262
 263             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 264
 265             /* Evaluate switch function */
 266             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 267             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 268             velec            = _mm_mul_ps(velec,sw);
 269             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 270
 271             /* Update potential sum for this i atom from the interaction with this j atom. */
 272             velec            = _mm_and_ps(velec,cutoff_mask);
 273             velecsum         = _mm_add_ps(velecsum,velec);
 274
 275             fscal            = felec;
 276
 277             fscal            = _mm_and_ps(fscal,cutoff_mask);
 278
 279              /* Update vectorial force */
 280             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 281             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 282             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 283
 284             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 285             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 286             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 287
 288             }
 289
 290             /**************************
 291              * CALCULATE INTERACTIONS *
 292              **************************/
 293
 294             if (gmx_mm_any_lt(rsq20,rcutoff2))
 295             {
 296
 297             r20              = _mm_mul_ps(rsq20,rinv20);
 298
 299             /* Compute parameters for interactions between i and j atoms */
 300             qq20             = _mm_mul_ps(iq2,jq0);
 301
 302             /* EWALD ELECTROSTATICS */
 303
 304             /* Analytical PME correction */
 305             zeta2            = _mm_mul_ps(beta2,rsq20);
 306             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
 307             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 308             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 309             felec            = _mm_mul_ps(qq20,felec);
 310             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 311             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
 312             velec            = _mm_mul_ps(qq20,velec);
 313
 314             d                = _mm_sub_ps(r20,rswitch);
 315             d                = _mm_max_ps(d,_mm_setzero_ps());
 316             d2               = _mm_mul_ps(d,d);
 317             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 318
 319             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 320
 321             /* Evaluate switch function */
 322             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 323             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
 324             velec            = _mm_mul_ps(velec,sw);
 325             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 326
 327             /* Update potential sum for this i atom from the interaction with this j atom. */
 328             velec            = _mm_and_ps(velec,cutoff_mask);
 329             velecsum         = _mm_add_ps(velecsum,velec);
 330
 331             fscal            = felec;
 332
 333             fscal            = _mm_and_ps(fscal,cutoff_mask);
 334
 335              /* Update vectorial force */
 336             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 337             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 338             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 339
 340             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 341             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 342             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 343
 344             }
 345
 346             /**************************
 347              * CALCULATE INTERACTIONS *
 348              **************************/
 349
 350             if (gmx_mm_any_lt(rsq30,rcutoff2))
 351             {
 352
 353             r30              = _mm_mul_ps(rsq30,rinv30);
 354
 355             /* Compute parameters for interactions between i and j atoms */
 356             qq30             = _mm_mul_ps(iq3,jq0);
 357
 358             /* EWALD ELECTROSTATICS */
 359
 360             /* Analytical PME correction */
 361             zeta2            = _mm_mul_ps(beta2,rsq30);
 362             rinv3            = _mm_mul_ps(rinvsq30,rinv30);
 363             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 364             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 365             felec            = _mm_mul_ps(qq30,felec);
 366             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 367             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv30);
 368             velec            = _mm_mul_ps(qq30,velec);
 369
 370             d                = _mm_sub_ps(r30,rswitch);
 371             d                = _mm_max_ps(d,_mm_setzero_ps());
 372             d2               = _mm_mul_ps(d,d);
 373             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 374
 375             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 376
 377             /* Evaluate switch function */
 378             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 379             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
 380             velec            = _mm_mul_ps(velec,sw);
 381             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
 382
 383             /* Update potential sum for this i atom from the interaction with this j atom. */
 384             velec            = _mm_and_ps(velec,cutoff_mask);
 385             velecsum         = _mm_add_ps(velecsum,velec);
 386
 387             fscal            = felec;
 388
 389             fscal            = _mm_and_ps(fscal,cutoff_mask);
 390
 391              /* Update vectorial force */
 392             fix3             = _mm_macc_ps(dx30,fscal,fix3);
 393             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
 394             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
 395
 396             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
 397             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
 398             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
 399
 400             }
 401
 402             fjptrA             = f+j_coord_offsetA;
 403             fjptrB             = f+j_coord_offsetB;
 404             fjptrC             = f+j_coord_offsetC;
 405             fjptrD             = f+j_coord_offsetD;
 406
 407             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 408
 409             /* Inner loop uses 159 flops */
 410         }
 411
 412         if(jidx<j_index_end)
 413         {
 414
 415             /* Get j neighbor index, and coordinate index */
 416             jnrlistA         = jjnr[jidx];
 417             jnrlistB         = jjnr[jidx+1];
 418             jnrlistC         = jjnr[jidx+2];
 419             jnrlistD         = jjnr[jidx+3];
 420             /* Sign of each element will be negative for non-real atoms.
 421              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
 422              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
 423              */
 424             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
 425             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
 426             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
 427             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
 428             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
 429             j_coord_offsetA  = DIM*jnrA;
 430             j_coord_offsetB  = DIM*jnrB;
 431             j_coord_offsetC  = DIM*jnrC;
 432             j_coord_offsetD  = DIM*jnrD;
 433
 434             /* load j atom coordinates */
 435             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 436                                               x+j_coord_offsetC,x+j_coord_offsetD,
 437                                               &jx0,&jy0,&jz0);
 438
 439             /* Calculate displacement vector */
 440             dx10             = _mm_sub_ps(ix1,jx0);
 441             dy10             = _mm_sub_ps(iy1,jy0);
 442             dz10             = _mm_sub_ps(iz1,jz0);
 443             dx20             = _mm_sub_ps(ix2,jx0);
 444             dy20             = _mm_sub_ps(iy2,jy0);
 445             dz20             = _mm_sub_ps(iz2,jz0);
 446             dx30             = _mm_sub_ps(ix3,jx0);
 447             dy30             = _mm_sub_ps(iy3,jy0);
 448             dz30             = _mm_sub_ps(iz3,jz0);
 449
 450             /* Calculate squared distance and things based on it */
 451             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 452             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 453             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 454
 455             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 456             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 457             rinv30           = gmx_mm_invsqrt_ps(rsq30);
 458
 459             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 460             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 461             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 462
 463             /* Load parameters for j particles */
 464             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 465                                                               charge+jnrC+0,charge+jnrD+0);
 466
 467             fjx0             = _mm_setzero_ps();
 468             fjy0             = _mm_setzero_ps();
 469             fjz0             = _mm_setzero_ps();
 470
 471             /**************************
 472              * CALCULATE INTERACTIONS *
 473              **************************/
 474
 475             if (gmx_mm_any_lt(rsq10,rcutoff2))
 476             {
 477
 478             r10              = _mm_mul_ps(rsq10,rinv10);
 479             r10              = _mm_andnot_ps(dummy_mask,r10);
 480
 481             /* Compute parameters for interactions between i and j atoms */
 482             qq10             = _mm_mul_ps(iq1,jq0);
 483
 484             /* EWALD ELECTROSTATICS */
 485
 486             /* Analytical PME correction */
 487             zeta2            = _mm_mul_ps(beta2,rsq10);
 488             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 489             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 490             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 491             felec            = _mm_mul_ps(qq10,felec);
 492             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 493             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 494             velec            = _mm_mul_ps(qq10,velec);
 495
 496             d                = _mm_sub_ps(r10,rswitch);
 497             d                = _mm_max_ps(d,_mm_setzero_ps());
 498             d2               = _mm_mul_ps(d,d);
 499             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 500
 501             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 502
 503             /* Evaluate switch function */
 504             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 505             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 506             velec            = _mm_mul_ps(velec,sw);
 507             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 508
 509             /* Update potential sum for this i atom from the interaction with this j atom. */
 510             velec            = _mm_and_ps(velec,cutoff_mask);
 511             velec            = _mm_andnot_ps(dummy_mask,velec);
 512             velecsum         = _mm_add_ps(velecsum,velec);
 513
 514             fscal            = felec;
 515
 516             fscal            = _mm_and_ps(fscal,cutoff_mask);
 517
 518             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 519
 520              /* Update vectorial force */
 521             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 522             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 523             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 524
 525             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 526             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 527             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 528
 529             }
 530
 531             /**************************
 532              * CALCULATE INTERACTIONS *
 533              **************************/
 534
 535             if (gmx_mm_any_lt(rsq20,rcutoff2))
 536             {
 537
 538             r20              = _mm_mul_ps(rsq20,rinv20);
 539             r20              = _mm_andnot_ps(dummy_mask,r20);
 540
 541             /* Compute parameters for interactions between i and j atoms */
 542             qq20             = _mm_mul_ps(iq2,jq0);
 543
 544             /* EWALD ELECTROSTATICS */
 545
 546             /* Analytical PME correction */
 547             zeta2            = _mm_mul_ps(beta2,rsq20);
 548             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
 549             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 550             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 551             felec            = _mm_mul_ps(qq20,felec);
 552             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 553             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
 554             velec            = _mm_mul_ps(qq20,velec);
 555
 556             d                = _mm_sub_ps(r20,rswitch);
 557             d                = _mm_max_ps(d,_mm_setzero_ps());
 558             d2               = _mm_mul_ps(d,d);
 559             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 560
 561             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 562
 563             /* Evaluate switch function */
 564             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 565             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
 566             velec            = _mm_mul_ps(velec,sw);
 567             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 568
 569             /* Update potential sum for this i atom from the interaction with this j atom. */
 570             velec            = _mm_and_ps(velec,cutoff_mask);
 571             velec            = _mm_andnot_ps(dummy_mask,velec);
 572             velecsum         = _mm_add_ps(velecsum,velec);
 573
 574             fscal            = felec;
 575
 576             fscal            = _mm_and_ps(fscal,cutoff_mask);
 577
 578             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 579
 580              /* Update vectorial force */
 581             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 582             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 583             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 584
 585             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 586             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 587             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 588
 589             }
 590
 591             /**************************
 592              * CALCULATE INTERACTIONS *
 593              **************************/
 594
 595             if (gmx_mm_any_lt(rsq30,rcutoff2))
 596             {
 597
 598             r30              = _mm_mul_ps(rsq30,rinv30);
 599             r30              = _mm_andnot_ps(dummy_mask,r30);
 600
 601             /* Compute parameters for interactions between i and j atoms */
 602             qq30             = _mm_mul_ps(iq3,jq0);
 603
 604             /* EWALD ELECTROSTATICS */
 605
 606             /* Analytical PME correction */
 607             zeta2            = _mm_mul_ps(beta2,rsq30);
 608             rinv3            = _mm_mul_ps(rinvsq30,rinv30);
 609             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 610             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 611             felec            = _mm_mul_ps(qq30,felec);
 612             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 613             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv30);
 614             velec            = _mm_mul_ps(qq30,velec);
 615
 616             d                = _mm_sub_ps(r30,rswitch);
 617             d                = _mm_max_ps(d,_mm_setzero_ps());
 618             d2               = _mm_mul_ps(d,d);
 619             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 620
 621             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 622
 623             /* Evaluate switch function */
 624             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 625             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
 626             velec            = _mm_mul_ps(velec,sw);
 627             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
 628
 629             /* Update potential sum for this i atom from the interaction with this j atom. */
 630             velec            = _mm_and_ps(velec,cutoff_mask);
 631             velec            = _mm_andnot_ps(dummy_mask,velec);
 632             velecsum         = _mm_add_ps(velecsum,velec);
 633
 634             fscal            = felec;
 635
 636             fscal            = _mm_and_ps(fscal,cutoff_mask);
 637
 638             fscal            = _mm_andnot_ps(dummy_mask,fscal);
 639
 640              /* Update vectorial force */
 641             fix3             = _mm_macc_ps(dx30,fscal,fix3);
 642             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
 643             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
 644
 645             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
 646             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
 647             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
 648
 649             }
 650
 651             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
 652             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
 653             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
 654             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
 655
 656             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
 657
 658             /* Inner loop uses 162 flops */
 659         }
 660
 661         /* End of innermost loop */
 662
 663         gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
 664                                               f+i_coord_offset+DIM,fshift+i_shift_offset);
 665
 666         ggid                        = gid[iidx];
 667         /* Update potential energies */
 668         gmx_mm_update_1pot_ps(velecsum,kernel_data->energygrp_elec+ggid);
 669
 670         /* Increment number of inner iterations */
 671         inneriter                  += j_index_end - j_index_start;
 672
 673         /* Outer loop uses 19 flops */
 674     }
 675
 676     /* Increment number of outer iterations */
 677     outeriter        += nri;
 678
 679     /* Update outer/inner flops */
 680
 681     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_VF,outeriter*19 + inneriter*162);
 682 }
 683 /*
 684  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwNone_GeomW4P1_F_avx_128_fma_single
 685  * Electrostatics interaction: Ewald
 686  * VdW interaction:            None
 687  * Geometry:                   Water4-Particle
 688  * Calculate force/pot:        Force
 689  */
 690 void
 691 nb_kernel_ElecEwSw_VdwNone_GeomW4P1_F_avx_128_fma_single
 692                     (t_nblist * gmx_restrict                nlist,
 693                      rvec * gmx_restrict                    xx,
 694                      rvec * gmx_restrict                    ff,
 695                      t_forcerec * gmx_restrict              fr,
 696                      t_mdatoms * gmx_restrict               mdatoms,
 697                      nb_kernel_data_t * gmx_restrict        kernel_data,
 698                      t_nrnb * gmx_restrict                  nrnb)
 699 {
 700     /* Suffixes 0,1,2,3 refer to particle indices for waters in the inner or outer loop, or
 701      * just 0 for non-waters.
 702      * Suffixes A,B,C,D refer to j loop unrolling done with AVX_128, e.g. for the four different
 703      * jnr indices corresponding to data put in the four positions in the SIMD register.
 704      */
 705     int              i_shift_offset,i_coord_offset,outeriter,inneriter;
 706     int              j_index_start,j_index_end,jidx,nri,inr,ggid,iidx;
 707     int              jnrA,jnrB,jnrC,jnrD;
 708     int              jnrlistA,jnrlistB,jnrlistC,jnrlistD;
 709     int              j_coord_offsetA,j_coord_offsetB,j_coord_offsetC,j_coord_offsetD;
 710     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 711     real             rcutoff_scalar;
 712     real             *shiftvec,*fshift,*x,*f;
 713     real             *fjptrA,*fjptrB,*fjptrC,*fjptrD;
 714     real             scratch[4*DIM];
 715     __m128           fscal,rcutoff,rcutoff2,jidxall;
 716     int              vdwioffset1;
 717     __m128           ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 718     int              vdwioffset2;
 719     __m128           ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 720     int              vdwioffset3;
 721     __m128           ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 722     int              vdwjidx0A,vdwjidx0B,vdwjidx0C,vdwjidx0D;
 723     __m128           jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 724     __m128           dx10,dy10,dz10,rsq10,rinv10,rinvsq10,r10,qq10,c6_10,c12_10;
 725     __m128           dx20,dy20,dz20,rsq20,rinv20,rinvsq20,r20,qq20,c6_20,c12_20;
 726     __m128           dx30,dy30,dz30,rsq30,rinv30,rinvsq30,r30,qq30,c6_30,c12_30;
 727     __m128           velec,felec,velecsum,facel,crf,krf,krf2;
 728     real             *charge;
 729     __m128i          ewitab;
 730     __m128           ewtabscale,eweps,twoeweps,sh_ewald,ewrt,ewtabhalfspace,ewtabF,ewtabFn,ewtabD,ewtabV;
 731     __m128           beta,beta2,beta3,zeta2,pmecorrF,pmecorrV,rinv3;
 732     real             *ewtab;
 733     __m128           rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 734     real             rswitch_scalar,d_scalar;
 735     __m128           dummy_mask,cutoff_mask;
 736     __m128           signbit = _mm_castsi128_ps( _mm_set1_epi32(0x80000000) );
 737     __m128           one     = _mm_set1_ps(1.0);
 738     __m128           two     = _mm_set1_ps(2.0);
 739     x                = xx[0];
 740     f                = ff[0];
 741
 742     nri              = nlist->nri;
 743     iinr             = nlist->iinr;
 744     jindex           = nlist->jindex;
 745     jjnr             = nlist->jjnr;
 746     shiftidx         = nlist->shift;
 747     gid              = nlist->gid;
 748     shiftvec         = fr->shift_vec[0];
 749     fshift           = fr->fshift[0];
 750     facel            = _mm_set1_ps(fr->epsfac);
 751     charge           = mdatoms->chargeA;
 752
 753     sh_ewald         = _mm_set1_ps(fr->ic->sh_ewald);
 754     beta             = _mm_set1_ps(fr->ic->ewaldcoeff);
 755     beta2            = _mm_mul_ps(beta,beta);
 756     beta3            = _mm_mul_ps(beta,beta2);
 757     ewtab            = fr->ic->tabq_coul_FDV0;
 758     ewtabscale       = _mm_set1_ps(fr->ic->tabq_scale);
 759     ewtabhalfspace   = _mm_set1_ps(0.5/fr->ic->tabq_scale);
 760
 761     /* Setup water-specific parameters */
 762     inr              = nlist->iinr[0];
 763     iq1              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+1]));
 764     iq2              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+2]));
 765     iq3              = _mm_mul_ps(facel,_mm_set1_ps(charge[inr+3]));
 766
 767     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 768     rcutoff_scalar   = fr->rcoulomb;
 769     rcutoff          = _mm_set1_ps(rcutoff_scalar);
 770     rcutoff2         = _mm_mul_ps(rcutoff,rcutoff);
 771
 772     rswitch_scalar   = fr->rcoulomb_switch;
 773     rswitch          = _mm_set1_ps(rswitch_scalar);
 774     /* Setup switch parameters */
 775     d_scalar         = rcutoff_scalar-rswitch_scalar;
 776     d                = _mm_set1_ps(d_scalar);
 777     swV3             = _mm_set1_ps(-10.0/(d_scalar*d_scalar*d_scalar));
 778     swV4             = _mm_set1_ps( 15.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 779     swV5             = _mm_set1_ps( -6.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 780     swF2             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar));
 781     swF3             = _mm_set1_ps( 60.0/(d_scalar*d_scalar*d_scalar*d_scalar));
 782     swF4             = _mm_set1_ps(-30.0/(d_scalar*d_scalar*d_scalar*d_scalar*d_scalar));
 783
 784     /* Avoid stupid compiler warnings */
 785     jnrA = jnrB = jnrC = jnrD = 0;
 786     j_coord_offsetA = 0;
 787     j_coord_offsetB = 0;
 788     j_coord_offsetC = 0;
 789     j_coord_offsetD = 0;
 790
 791     outeriter        = 0;
 792     inneriter        = 0;
 793
 794     for(iidx=0;iidx<4*DIM;iidx++)
 795     {
 796         scratch[iidx] = 0.0;
 797     }
 798
 799     /* Start outer loop over neighborlists */
 800     for(iidx=0; iidx<nri; iidx++)
 801     {
 802         /* Load shift vector for this list */
 803         i_shift_offset   = DIM*shiftidx[iidx];
 804
 805         /* Load limits for loop over neighbors */
 806         j_index_start    = jindex[iidx];
 807         j_index_end      = jindex[iidx+1];
 808
 809         /* Get outer coordinate index */
 810         inr              = iinr[iidx];
 811         i_coord_offset   = DIM*inr;
 812
 813         /* Load i particle coords and add shift vector */
 814         gmx_mm_load_shift_and_3rvec_broadcast_ps(shiftvec+i_shift_offset,x+i_coord_offset+DIM,
 815                                                  &ix1,&iy1,&iz1,&ix2,&iy2,&iz2,&ix3,&iy3,&iz3);
 816
 817         fix1             = _mm_setzero_ps();
 818         fiy1             = _mm_setzero_ps();
 819         fiz1             = _mm_setzero_ps();
 820         fix2             = _mm_setzero_ps();
 821         fiy2             = _mm_setzero_ps();
 822         fiz2             = _mm_setzero_ps();
 823         fix3             = _mm_setzero_ps();
 824         fiy3             = _mm_setzero_ps();
 825         fiz3             = _mm_setzero_ps();
 826
 827         /* Start inner kernel loop */
 828         for(jidx=j_index_start; jidx<j_index_end && jjnr[jidx+3]>=0; jidx+=4)
 829         {
 830
 831             /* Get j neighbor index, and coordinate index */
 832             jnrA             = jjnr[jidx];
 833             jnrB             = jjnr[jidx+1];
 834             jnrC             = jjnr[jidx+2];
 835             jnrD             = jjnr[jidx+3];
 836             j_coord_offsetA  = DIM*jnrA;
 837             j_coord_offsetB  = DIM*jnrB;
 838             j_coord_offsetC  = DIM*jnrC;
 839             j_coord_offsetD  = DIM*jnrD;
 840
 841             /* load j atom coordinates */
 842             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
 843                                               x+j_coord_offsetC,x+j_coord_offsetD,
 844                                               &jx0,&jy0,&jz0);
 845
 846             /* Calculate displacement vector */
 847             dx10             = _mm_sub_ps(ix1,jx0);
 848             dy10             = _mm_sub_ps(iy1,jy0);
 849             dz10             = _mm_sub_ps(iz1,jz0);
 850             dx20             = _mm_sub_ps(ix2,jx0);
 851             dy20             = _mm_sub_ps(iy2,jy0);
 852             dz20             = _mm_sub_ps(iz2,jz0);
 853             dx30             = _mm_sub_ps(ix3,jx0);
 854             dy30             = _mm_sub_ps(iy3,jy0);
 855             dz30             = _mm_sub_ps(iz3,jz0);
 856
 857             /* Calculate squared distance and things based on it */
 858             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
 859             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
 860             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
 861
 862             rinv10           = gmx_mm_invsqrt_ps(rsq10);
 863             rinv20           = gmx_mm_invsqrt_ps(rsq20);
 864             rinv30           = gmx_mm_invsqrt_ps(rsq30);
 865
 866             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
 867             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
 868             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
 869
 870             /* Load parameters for j particles */
 871             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
 872                                                               charge+jnrC+0,charge+jnrD+0);
 873
 874             fjx0             = _mm_setzero_ps();
 875             fjy0             = _mm_setzero_ps();
 876             fjz0             = _mm_setzero_ps();
 877
 878             /**************************
 879              * CALCULATE INTERACTIONS *
 880              **************************/
 881
 882             if (gmx_mm_any_lt(rsq10,rcutoff2))
 883             {
 884
 885             r10              = _mm_mul_ps(rsq10,rinv10);
 886
 887             /* Compute parameters for interactions between i and j atoms */
 888             qq10             = _mm_mul_ps(iq1,jq0);
 889
 890             /* EWALD ELECTROSTATICS */
 891
 892             /* Analytical PME correction */
 893             zeta2            = _mm_mul_ps(beta2,rsq10);
 894             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
 895             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 896             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 897             felec            = _mm_mul_ps(qq10,felec);
 898             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 899             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
 900             velec            = _mm_mul_ps(qq10,velec);
 901
 902             d                = _mm_sub_ps(r10,rswitch);
 903             d                = _mm_max_ps(d,_mm_setzero_ps());
 904             d2               = _mm_mul_ps(d,d);
 905             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 906
 907             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 908
 909             /* Evaluate switch function */
 910             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 911             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
 912             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
 913
 914             fscal            = felec;
 915
 916             fscal            = _mm_and_ps(fscal,cutoff_mask);
 917
 918              /* Update vectorial force */
 919             fix1             = _mm_macc_ps(dx10,fscal,fix1);
 920             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
 921             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
 922
 923             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
 924             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
 925             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
 926
 927             }
 928
 929             /**************************
 930              * CALCULATE INTERACTIONS *
 931              **************************/
 932
 933             if (gmx_mm_any_lt(rsq20,rcutoff2))
 934             {
 935
 936             r20              = _mm_mul_ps(rsq20,rinv20);
 937
 938             /* Compute parameters for interactions between i and j atoms */
 939             qq20             = _mm_mul_ps(iq2,jq0);
 940
 941             /* EWALD ELECTROSTATICS */
 942
 943             /* Analytical PME correction */
 944             zeta2            = _mm_mul_ps(beta2,rsq20);
 945             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
 946             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 947             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 948             felec            = _mm_mul_ps(qq20,felec);
 949             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
 950             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
 951             velec            = _mm_mul_ps(qq20,velec);
 952
 953             d                = _mm_sub_ps(r20,rswitch);
 954             d                = _mm_max_ps(d,_mm_setzero_ps());
 955             d2               = _mm_mul_ps(d,d);
 956             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
 957
 958             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
 959
 960             /* Evaluate switch function */
 961             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 962             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
 963             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
 964
 965             fscal            = felec;
 966
 967             fscal            = _mm_and_ps(fscal,cutoff_mask);
 968
 969              /* Update vectorial force */
 970             fix2             = _mm_macc_ps(dx20,fscal,fix2);
 971             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
 972             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
 973
 974             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
 975             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
 976             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
 977
 978             }
 979
 980             /**************************
 981              * CALCULATE INTERACTIONS *
 982              **************************/
 983
 984             if (gmx_mm_any_lt(rsq30,rcutoff2))
 985             {
 986
 987             r30              = _mm_mul_ps(rsq30,rinv30);
 988
 989             /* Compute parameters for interactions between i and j atoms */
 990             qq30             = _mm_mul_ps(iq3,jq0);
 991
 992             /* EWALD ELECTROSTATICS */
 993
 994             /* Analytical PME correction */
 995             zeta2            = _mm_mul_ps(beta2,rsq30);
 996             rinv3            = _mm_mul_ps(rinvsq30,rinv30);
 997             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
 998             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
 999             felec            = _mm_mul_ps(qq30,felec);
1000             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1001             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv30);
1002             velec            = _mm_mul_ps(qq30,velec);
1003
1004             d                = _mm_sub_ps(r30,rswitch);
1005             d                = _mm_max_ps(d,_mm_setzero_ps());
1006             d2               = _mm_mul_ps(d,d);
1007             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1008
1009             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1010
1011             /* Evaluate switch function */
1012             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1013             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
1014             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
1015
1016             fscal            = felec;
1017
1018             fscal            = _mm_and_ps(fscal,cutoff_mask);
1019
1020              /* Update vectorial force */
1021             fix3             = _mm_macc_ps(dx30,fscal,fix3);
1022             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
1023             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
1024
1025             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
1026             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
1027             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
1028
1029             }
1030
1031             fjptrA             = f+j_coord_offsetA;
1032             fjptrB             = f+j_coord_offsetB;
1033             fjptrC             = f+j_coord_offsetC;
1034             fjptrD             = f+j_coord_offsetD;
1035
1036             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1037
1038             /* Inner loop uses 150 flops */
1039         }
1040
1041         if(jidx<j_index_end)
1042         {
1043
1044             /* Get j neighbor index, and coordinate index */
1045             jnrlistA         = jjnr[jidx];
1046             jnrlistB         = jjnr[jidx+1];
1047             jnrlistC         = jjnr[jidx+2];
1048             jnrlistD         = jjnr[jidx+3];
1049             /* Sign of each element will be negative for non-real atoms.
1050              * This mask will be 0xFFFFFFFF for dummy entries and 0x0 for real ones,
1051              * so use it as val = _mm_andnot_ps(mask,val) to clear dummy entries.
1052              */
1053             dummy_mask = gmx_mm_castsi128_ps(_mm_cmplt_epi32(_mm_loadu_si128((const __m128i *)(jjnr+jidx)),_mm_setzero_si128()));
1054             jnrA       = (jnrlistA>=0) ? jnrlistA : 0;
1055             jnrB       = (jnrlistB>=0) ? jnrlistB : 0;
1056             jnrC       = (jnrlistC>=0) ? jnrlistC : 0;
1057             jnrD       = (jnrlistD>=0) ? jnrlistD : 0;
1058             j_coord_offsetA  = DIM*jnrA;
1059             j_coord_offsetB  = DIM*jnrB;
1060             j_coord_offsetC  = DIM*jnrC;
1061             j_coord_offsetD  = DIM*jnrD;
1062
1063             /* load j atom coordinates */
1064             gmx_mm_load_1rvec_4ptr_swizzle_ps(x+j_coord_offsetA,x+j_coord_offsetB,
1065                                               x+j_coord_offsetC,x+j_coord_offsetD,
1066                                               &jx0,&jy0,&jz0);
1067
1068             /* Calculate displacement vector */
1069             dx10             = _mm_sub_ps(ix1,jx0);
1070             dy10             = _mm_sub_ps(iy1,jy0);
1071             dz10             = _mm_sub_ps(iz1,jz0);
1072             dx20             = _mm_sub_ps(ix2,jx0);
1073             dy20             = _mm_sub_ps(iy2,jy0);
1074             dz20             = _mm_sub_ps(iz2,jz0);
1075             dx30             = _mm_sub_ps(ix3,jx0);
1076             dy30             = _mm_sub_ps(iy3,jy0);
1077             dz30             = _mm_sub_ps(iz3,jz0);
1078
1079             /* Calculate squared distance and things based on it */
1080             rsq10            = gmx_mm_calc_rsq_ps(dx10,dy10,dz10);
1081             rsq20            = gmx_mm_calc_rsq_ps(dx20,dy20,dz20);
1082             rsq30            = gmx_mm_calc_rsq_ps(dx30,dy30,dz30);
1083
1084             rinv10           = gmx_mm_invsqrt_ps(rsq10);
1085             rinv20           = gmx_mm_invsqrt_ps(rsq20);
1086             rinv30           = gmx_mm_invsqrt_ps(rsq30);
1087
1088             rinvsq10         = _mm_mul_ps(rinv10,rinv10);
1089             rinvsq20         = _mm_mul_ps(rinv20,rinv20);
1090             rinvsq30         = _mm_mul_ps(rinv30,rinv30);
1091
1092             /* Load parameters for j particles */
1093             jq0              = gmx_mm_load_4real_swizzle_ps(charge+jnrA+0,charge+jnrB+0,
1094                                                               charge+jnrC+0,charge+jnrD+0);
1095
1096             fjx0             = _mm_setzero_ps();
1097             fjy0             = _mm_setzero_ps();
1098             fjz0             = _mm_setzero_ps();
1099
1100             /**************************
1101              * CALCULATE INTERACTIONS *
1102              **************************/
1103
1104             if (gmx_mm_any_lt(rsq10,rcutoff2))
1105             {
1106
1107             r10              = _mm_mul_ps(rsq10,rinv10);
1108             r10              = _mm_andnot_ps(dummy_mask,r10);
1109
1110             /* Compute parameters for interactions between i and j atoms */
1111             qq10             = _mm_mul_ps(iq1,jq0);
1112
1113             /* EWALD ELECTROSTATICS */
1114
1115             /* Analytical PME correction */
1116             zeta2            = _mm_mul_ps(beta2,rsq10);
1117             rinv3            = _mm_mul_ps(rinvsq10,rinv10);
1118             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1119             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1120             felec            = _mm_mul_ps(qq10,felec);
1121             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1122             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv10);
1123             velec            = _mm_mul_ps(qq10,velec);
1124
1125             d                = _mm_sub_ps(r10,rswitch);
1126             d                = _mm_max_ps(d,_mm_setzero_ps());
1127             d2               = _mm_mul_ps(d,d);
1128             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1129
1130             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1131
1132             /* Evaluate switch function */
1133             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1134             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv10,_mm_mul_ps(velec,dsw)) );
1135             cutoff_mask      = _mm_cmplt_ps(rsq10,rcutoff2);
1136
1137             fscal            = felec;
1138
1139             fscal            = _mm_and_ps(fscal,cutoff_mask);
1140
1141             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1142
1143              /* Update vectorial force */
1144             fix1             = _mm_macc_ps(dx10,fscal,fix1);
1145             fiy1             = _mm_macc_ps(dy10,fscal,fiy1);
1146             fiz1             = _mm_macc_ps(dz10,fscal,fiz1);
1147
1148             fjx0             = _mm_macc_ps(dx10,fscal,fjx0);
1149             fjy0             = _mm_macc_ps(dy10,fscal,fjy0);
1150             fjz0             = _mm_macc_ps(dz10,fscal,fjz0);
1151
1152             }
1153
1154             /**************************
1155              * CALCULATE INTERACTIONS *
1156              **************************/
1157
1158             if (gmx_mm_any_lt(rsq20,rcutoff2))
1159             {
1160
1161             r20              = _mm_mul_ps(rsq20,rinv20);
1162             r20              = _mm_andnot_ps(dummy_mask,r20);
1163
1164             /* Compute parameters for interactions between i and j atoms */
1165             qq20             = _mm_mul_ps(iq2,jq0);
1166
1167             /* EWALD ELECTROSTATICS */
1168
1169             /* Analytical PME correction */
1170             zeta2            = _mm_mul_ps(beta2,rsq20);
1171             rinv3            = _mm_mul_ps(rinvsq20,rinv20);
1172             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1173             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1174             felec            = _mm_mul_ps(qq20,felec);
1175             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1176             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv20);
1177             velec            = _mm_mul_ps(qq20,velec);
1178
1179             d                = _mm_sub_ps(r20,rswitch);
1180             d                = _mm_max_ps(d,_mm_setzero_ps());
1181             d2               = _mm_mul_ps(d,d);
1182             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1183
1184             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1185
1186             /* Evaluate switch function */
1187             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1188             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv20,_mm_mul_ps(velec,dsw)) );
1189             cutoff_mask      = _mm_cmplt_ps(rsq20,rcutoff2);
1190
1191             fscal            = felec;
1192
1193             fscal            = _mm_and_ps(fscal,cutoff_mask);
1194
1195             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1196
1197              /* Update vectorial force */
1198             fix2             = _mm_macc_ps(dx20,fscal,fix2);
1199             fiy2             = _mm_macc_ps(dy20,fscal,fiy2);
1200             fiz2             = _mm_macc_ps(dz20,fscal,fiz2);
1201
1202             fjx0             = _mm_macc_ps(dx20,fscal,fjx0);
1203             fjy0             = _mm_macc_ps(dy20,fscal,fjy0);
1204             fjz0             = _mm_macc_ps(dz20,fscal,fjz0);
1205
1206             }
1207
1208             /**************************
1209              * CALCULATE INTERACTIONS *
1210              **************************/
1211
1212             if (gmx_mm_any_lt(rsq30,rcutoff2))
1213             {
1214
1215             r30              = _mm_mul_ps(rsq30,rinv30);
1216             r30              = _mm_andnot_ps(dummy_mask,r30);
1217
1218             /* Compute parameters for interactions between i and j atoms */
1219             qq30             = _mm_mul_ps(iq3,jq0);
1220
1221             /* EWALD ELECTROSTATICS */
1222
1223             /* Analytical PME correction */
1224             zeta2            = _mm_mul_ps(beta2,rsq30);
1225             rinv3            = _mm_mul_ps(rinvsq30,rinv30);
1226             pmecorrF         = gmx_mm_pmecorrF_ps(zeta2);
1227             felec            = _mm_macc_ps(pmecorrF,beta3,rinv3);
1228             felec            = _mm_mul_ps(qq30,felec);
1229             pmecorrV         = gmx_mm_pmecorrV_ps(zeta2);
1230             velec            = _mm_nmacc_ps(pmecorrV,beta,rinv30);
1231             velec            = _mm_mul_ps(qq30,velec);
1232
1233             d                = _mm_sub_ps(r30,rswitch);
1234             d                = _mm_max_ps(d,_mm_setzero_ps());
1235             d2               = _mm_mul_ps(d,d);
1236             sw               = _mm_add_ps(one,_mm_mul_ps(d2,_mm_mul_ps(d,_mm_macc_ps(d,_mm_macc_ps(d,swV5,swV4),swV3))));
1237
1238             dsw              = _mm_mul_ps(d2,_mm_macc_ps(d,_mm_macc_ps(d,swF4,swF3),swF2));
1239
1240             /* Evaluate switch function */
1241             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1242             felec            = _mm_msub_ps( felec,sw , _mm_mul_ps(rinv30,_mm_mul_ps(velec,dsw)) );
1243             cutoff_mask      = _mm_cmplt_ps(rsq30,rcutoff2);
1244
1245             fscal            = felec;
1246
1247             fscal            = _mm_and_ps(fscal,cutoff_mask);
1248
1249             fscal            = _mm_andnot_ps(dummy_mask,fscal);
1250
1251              /* Update vectorial force */
1252             fix3             = _mm_macc_ps(dx30,fscal,fix3);
1253             fiy3             = _mm_macc_ps(dy30,fscal,fiy3);
1254             fiz3             = _mm_macc_ps(dz30,fscal,fiz3);
1255
1256             fjx0             = _mm_macc_ps(dx30,fscal,fjx0);
1257             fjy0             = _mm_macc_ps(dy30,fscal,fjy0);
1258             fjz0             = _mm_macc_ps(dz30,fscal,fjz0);
1259
1260             }
1261
1262             fjptrA             = (jnrlistA>=0) ? f+j_coord_offsetA : scratch;
1263             fjptrB             = (jnrlistB>=0) ? f+j_coord_offsetB : scratch;
1264             fjptrC             = (jnrlistC>=0) ? f+j_coord_offsetC : scratch;
1265             fjptrD             = (jnrlistD>=0) ? f+j_coord_offsetD : scratch;
1266
1267             gmx_mm_decrement_1rvec_4ptr_swizzle_ps(fjptrA,fjptrB,fjptrC,fjptrD,fjx0,fjy0,fjz0);
1268
1269             /* Inner loop uses 153 flops */
1270         }
1271
1272         /* End of innermost loop */
1273
1274         gmx_mm_update_iforce_3atom_swizzle_ps(fix1,fiy1,fiz1,fix2,fiy2,fiz2,fix3,fiy3,fiz3,
1275                                               f+i_coord_offset+DIM,fshift+i_shift_offset);
1276
1277         /* Increment number of inner iterations */
1278         inneriter                  += j_index_end - j_index_start;
1279
1280         /* Outer loop uses 18 flops */
1281     }
1282
1283     /* Increment number of outer iterations */
1284     outeriter        += nri;
1285
1286     /* Update outer/inner flops */
1287
1288     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_W4_F,outeriter*18 + inneriter*153);
1289 }