src/gmxlib/nonbonded/nb_kernel_c/nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_c.c

   1 /*
   2  * Note: this file was generated by the Gromacs c 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 /*
  34  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_VF_c
  35  * Electrostatics interaction: Ewald
  36  * VdW interaction:            Buckingham
  37  * Geometry:                   Water4-Water4
  38  * Calculate force/pot:        PotentialAndForce
  39  */
  40 void
  41 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_VF_c
  42                     (t_nblist * gmx_restrict                nlist,
  43                      rvec * gmx_restrict                    xx,
  44                      rvec * gmx_restrict                    ff,
  45                      t_forcerec * gmx_restrict              fr,
  46                      t_mdatoms * gmx_restrict               mdatoms,
  47                      nb_kernel_data_t * gmx_restrict        kernel_data,
  48                      t_nrnb * gmx_restrict                  nrnb)
  49 {
  50     int              i_shift_offset,i_coord_offset,j_coord_offset;
  51     int              j_index_start,j_index_end;
  52     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
  53     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
  54     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
  55     real             *shiftvec,*fshift,*x,*f;
  56     int              vdwioffset0;
  57     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
  58     int              vdwioffset1;
  59     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
  60     int              vdwioffset2;
  61     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
  62     int              vdwioffset3;
  63     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
  64     int              vdwjidx0;
  65     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
  66     int              vdwjidx1;
  67     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
  68     int              vdwjidx2;
  69     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
  70     int              vdwjidx3;
  71     real             jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
  72     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
  73     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
  74     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
  75     real             dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
  76     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
  77     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
  78     real             dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
  79     real             dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
  80     real             dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
  81     real             dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
  82     real             velec,felec,velecsum,facel,crf,krf,krf2;
  83     real             *charge;
  84     int              nvdwtype;
  85     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
  86     int              *vdwtype;
  87     real             *vdwparam;
  88     int              ewitab;
  89     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
  90     real             *ewtab;
  91     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
  92
  93     x                = xx[0];
  94     f                = ff[0];
  95
  96     nri              = nlist->nri;
  97     iinr             = nlist->iinr;
  98     jindex           = nlist->jindex;
  99     jjnr             = nlist->jjnr;
 100     shiftidx         = nlist->shift;
 101     gid              = nlist->gid;
 102     shiftvec         = fr->shift_vec[0];
 103     fshift           = fr->fshift[0];
 104     facel            = fr->epsfac;
 105     charge           = mdatoms->chargeA;
 106     nvdwtype         = fr->ntype;
 107     vdwparam         = fr->nbfp;
 108     vdwtype          = mdatoms->typeA;
 109
 110     sh_ewald         = fr->ic->sh_ewald;
 111     ewtab            = fr->ic->tabq_coul_FDV0;
 112     ewtabscale       = fr->ic->tabq_scale;
 113     ewtabhalfspace   = 0.5/ewtabscale;
 114
 115     /* Setup water-specific parameters */
 116     inr              = nlist->iinr[0];
 117     iq1              = facel*charge[inr+1];
 118     iq2              = facel*charge[inr+2];
 119     iq3              = facel*charge[inr+3];
 120     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 121
 122     jq1              = charge[inr+1];
 123     jq2              = charge[inr+2];
 124     jq3              = charge[inr+3];
 125     vdwjidx0         = 3*vdwtype[inr+0];
 126     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 127     cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 128     cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 129     qq11             = iq1*jq1;
 130     qq12             = iq1*jq2;
 131     qq13             = iq1*jq3;
 132     qq21             = iq2*jq1;
 133     qq22             = iq2*jq2;
 134     qq23             = iq2*jq3;
 135     qq31             = iq3*jq1;
 136     qq32             = iq3*jq2;
 137     qq33             = iq3*jq3;
 138
 139     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 140     rcutoff          = fr->rcoulomb;
 141     rcutoff2         = rcutoff*rcutoff;
 142
 143     rswitch          = fr->rcoulomb_switch;
 144     /* Setup switch parameters */
 145     d                = rcutoff-rswitch;
 146     swV3             = -10.0/(d*d*d);
 147     swV4             =  15.0/(d*d*d*d);
 148     swV5             =  -6.0/(d*d*d*d*d);
 149     swF2             = -30.0/(d*d*d);
 150     swF3             =  60.0/(d*d*d*d);
 151     swF4             = -30.0/(d*d*d*d*d);
 152
 153     outeriter        = 0;
 154     inneriter        = 0;
 155
 156     /* Start outer loop over neighborlists */
 157     for(iidx=0; iidx<nri; iidx++)
 158     {
 159         /* Load shift vector for this list */
 160         i_shift_offset   = DIM*shiftidx[iidx];
 161         shX              = shiftvec[i_shift_offset+XX];
 162         shY              = shiftvec[i_shift_offset+YY];
 163         shZ              = shiftvec[i_shift_offset+ZZ];
 164
 165         /* Load limits for loop over neighbors */
 166         j_index_start    = jindex[iidx];
 167         j_index_end      = jindex[iidx+1];
 168
 169         /* Get outer coordinate index */
 170         inr              = iinr[iidx];
 171         i_coord_offset   = DIM*inr;
 172
 173         /* Load i particle coords and add shift vector */
 174         ix0              = shX + x[i_coord_offset+DIM*0+XX];
 175         iy0              = shY + x[i_coord_offset+DIM*0+YY];
 176         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
 177         ix1              = shX + x[i_coord_offset+DIM*1+XX];
 178         iy1              = shY + x[i_coord_offset+DIM*1+YY];
 179         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
 180         ix2              = shX + x[i_coord_offset+DIM*2+XX];
 181         iy2              = shY + x[i_coord_offset+DIM*2+YY];
 182         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
 183         ix3              = shX + x[i_coord_offset+DIM*3+XX];
 184         iy3              = shY + x[i_coord_offset+DIM*3+YY];
 185         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
 186
 187         fix0             = 0.0;
 188         fiy0             = 0.0;
 189         fiz0             = 0.0;
 190         fix1             = 0.0;
 191         fiy1             = 0.0;
 192         fiz1             = 0.0;
 193         fix2             = 0.0;
 194         fiy2             = 0.0;
 195         fiz2             = 0.0;
 196         fix3             = 0.0;
 197         fiy3             = 0.0;
 198         fiz3             = 0.0;
 199
 200         /* Reset potential sums */
 201         velecsum         = 0.0;
 202         vvdwsum          = 0.0;
 203
 204         /* Start inner kernel loop */
 205         for(jidx=j_index_start; jidx<j_index_end; jidx++)
 206         {
 207             /* Get j neighbor index, and coordinate index */
 208             jnr              = jjnr[jidx];
 209             j_coord_offset   = DIM*jnr;
 210
 211             /* load j atom coordinates */
 212             jx0              = x[j_coord_offset+DIM*0+XX];
 213             jy0              = x[j_coord_offset+DIM*0+YY];
 214             jz0              = x[j_coord_offset+DIM*0+ZZ];
 215             jx1              = x[j_coord_offset+DIM*1+XX];
 216             jy1              = x[j_coord_offset+DIM*1+YY];
 217             jz1              = x[j_coord_offset+DIM*1+ZZ];
 218             jx2              = x[j_coord_offset+DIM*2+XX];
 219             jy2              = x[j_coord_offset+DIM*2+YY];
 220             jz2              = x[j_coord_offset+DIM*2+ZZ];
 221             jx3              = x[j_coord_offset+DIM*3+XX];
 222             jy3              = x[j_coord_offset+DIM*3+YY];
 223             jz3              = x[j_coord_offset+DIM*3+ZZ];
 224
 225             /* Calculate displacement vector */
 226             dx00             = ix0 - jx0;
 227             dy00             = iy0 - jy0;
 228             dz00             = iz0 - jz0;
 229             dx11             = ix1 - jx1;
 230             dy11             = iy1 - jy1;
 231             dz11             = iz1 - jz1;
 232             dx12             = ix1 - jx2;
 233             dy12             = iy1 - jy2;
 234             dz12             = iz1 - jz2;
 235             dx13             = ix1 - jx3;
 236             dy13             = iy1 - jy3;
 237             dz13             = iz1 - jz3;
 238             dx21             = ix2 - jx1;
 239             dy21             = iy2 - jy1;
 240             dz21             = iz2 - jz1;
 241             dx22             = ix2 - jx2;
 242             dy22             = iy2 - jy2;
 243             dz22             = iz2 - jz2;
 244             dx23             = ix2 - jx3;
 245             dy23             = iy2 - jy3;
 246             dz23             = iz2 - jz3;
 247             dx31             = ix3 - jx1;
 248             dy31             = iy3 - jy1;
 249             dz31             = iz3 - jz1;
 250             dx32             = ix3 - jx2;
 251             dy32             = iy3 - jy2;
 252             dz32             = iz3 - jz2;
 253             dx33             = ix3 - jx3;
 254             dy33             = iy3 - jy3;
 255             dz33             = iz3 - jz3;
 256
 257             /* Calculate squared distance and things based on it */
 258             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
 259             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 260             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
 261             rsq13            = dx13*dx13+dy13*dy13+dz13*dz13;
 262             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
 263             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
 264             rsq23            = dx23*dx23+dy23*dy23+dz23*dz23;
 265             rsq31            = dx31*dx31+dy31*dy31+dz31*dz31;
 266             rsq32            = dx32*dx32+dy32*dy32+dz32*dz32;
 267             rsq33            = dx33*dx33+dy33*dy33+dz33*dz33;
 268
 269             rinv00           = gmx_invsqrt(rsq00);
 270             rinv11           = gmx_invsqrt(rsq11);
 271             rinv12           = gmx_invsqrt(rsq12);
 272             rinv13           = gmx_invsqrt(rsq13);
 273             rinv21           = gmx_invsqrt(rsq21);
 274             rinv22           = gmx_invsqrt(rsq22);
 275             rinv23           = gmx_invsqrt(rsq23);
 276             rinv31           = gmx_invsqrt(rsq31);
 277             rinv32           = gmx_invsqrt(rsq32);
 278             rinv33           = gmx_invsqrt(rsq33);
 279
 280             rinvsq00         = rinv00*rinv00;
 281             rinvsq11         = rinv11*rinv11;
 282             rinvsq12         = rinv12*rinv12;
 283             rinvsq13         = rinv13*rinv13;
 284             rinvsq21         = rinv21*rinv21;
 285             rinvsq22         = rinv22*rinv22;
 286             rinvsq23         = rinv23*rinv23;
 287             rinvsq31         = rinv31*rinv31;
 288             rinvsq32         = rinv32*rinv32;
 289             rinvsq33         = rinv33*rinv33;
 290
 291             /**************************
 292              * CALCULATE INTERACTIONS *
 293              **************************/
 294
 295             if (rsq00<rcutoff2)
 296             {
 297
 298             r00              = rsq00*rinv00;
 299
 300             /* BUCKINGHAM DISPERSION/REPULSION */
 301             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
 302             vvdw6            = c6_00*rinvsix;
 303             br               = cexp2_00*r00;
 304             vvdwexp          = cexp1_00*exp(-br);
 305             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
 306             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
 307
 308             d                = r00-rswitch;
 309             d                = (d>0.0) ? d : 0.0;
 310             d2               = d*d;
 311             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 312
 313             dsw              = d2*(swF2+d*(swF3+d*swF4));
 314
 315             /* Evaluate switch function */
 316             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 317             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
 318             vvdw            *= sw;
 319
 320             /* Update potential sums from outer loop */
 321             vvdwsum         += vvdw;
 322
 323             fscal            = fvdw;
 324
 325             /* Calculate temporary vectorial force */
 326             tx               = fscal*dx00;
 327             ty               = fscal*dy00;
 328             tz               = fscal*dz00;
 329
 330             /* Update vectorial force */
 331             fix0            += tx;
 332             fiy0            += ty;
 333             fiz0            += tz;
 334             f[j_coord_offset+DIM*0+XX] -= tx;
 335             f[j_coord_offset+DIM*0+YY] -= ty;
 336             f[j_coord_offset+DIM*0+ZZ] -= tz;
 337
 338             }
 339
 340             /**************************
 341              * CALCULATE INTERACTIONS *
 342              **************************/
 343
 344             if (rsq11<rcutoff2)
 345             {
 346
 347             r11              = rsq11*rinv11;
 348
 349             /* EWALD ELECTROSTATICS */
 350
 351             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 352             ewrt             = r11*ewtabscale;
 353             ewitab           = ewrt;
 354             eweps            = ewrt-ewitab;
 355             ewitab           = 4*ewitab;
 356             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 357             velec            = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 358             felec            = qq11*rinv11*(rinvsq11-felec);
 359
 360             d                = r11-rswitch;
 361             d                = (d>0.0) ? d : 0.0;
 362             d2               = d*d;
 363             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 364
 365             dsw              = d2*(swF2+d*(swF3+d*swF4));
 366
 367             /* Evaluate switch function */
 368             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 369             felec            = felec*sw - rinv11*velec*dsw;
 370             velec           *= sw;
 371
 372             /* Update potential sums from outer loop */
 373             velecsum        += velec;
 374
 375             fscal            = felec;
 376
 377             /* Calculate temporary vectorial force */
 378             tx               = fscal*dx11;
 379             ty               = fscal*dy11;
 380             tz               = fscal*dz11;
 381
 382             /* Update vectorial force */
 383             fix1            += tx;
 384             fiy1            += ty;
 385             fiz1            += tz;
 386             f[j_coord_offset+DIM*1+XX] -= tx;
 387             f[j_coord_offset+DIM*1+YY] -= ty;
 388             f[j_coord_offset+DIM*1+ZZ] -= tz;
 389
 390             }
 391
 392             /**************************
 393              * CALCULATE INTERACTIONS *
 394              **************************/
 395
 396             if (rsq12<rcutoff2)
 397             {
 398
 399             r12              = rsq12*rinv12;
 400
 401             /* EWALD ELECTROSTATICS */
 402
 403             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 404             ewrt             = r12*ewtabscale;
 405             ewitab           = ewrt;
 406             eweps            = ewrt-ewitab;
 407             ewitab           = 4*ewitab;
 408             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 409             velec            = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 410             felec            = qq12*rinv12*(rinvsq12-felec);
 411
 412             d                = r12-rswitch;
 413             d                = (d>0.0) ? d : 0.0;
 414             d2               = d*d;
 415             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 416
 417             dsw              = d2*(swF2+d*(swF3+d*swF4));
 418
 419             /* Evaluate switch function */
 420             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 421             felec            = felec*sw - rinv12*velec*dsw;
 422             velec           *= sw;
 423
 424             /* Update potential sums from outer loop */
 425             velecsum        += velec;
 426
 427             fscal            = felec;
 428
 429             /* Calculate temporary vectorial force */
 430             tx               = fscal*dx12;
 431             ty               = fscal*dy12;
 432             tz               = fscal*dz12;
 433
 434             /* Update vectorial force */
 435             fix1            += tx;
 436             fiy1            += ty;
 437             fiz1            += tz;
 438             f[j_coord_offset+DIM*2+XX] -= tx;
 439             f[j_coord_offset+DIM*2+YY] -= ty;
 440             f[j_coord_offset+DIM*2+ZZ] -= tz;
 441
 442             }
 443
 444             /**************************
 445              * CALCULATE INTERACTIONS *
 446              **************************/
 447
 448             if (rsq13<rcutoff2)
 449             {
 450
 451             r13              = rsq13*rinv13;
 452
 453             /* EWALD ELECTROSTATICS */
 454
 455             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 456             ewrt             = r13*ewtabscale;
 457             ewitab           = ewrt;
 458             eweps            = ewrt-ewitab;
 459             ewitab           = 4*ewitab;
 460             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 461             velec            = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 462             felec            = qq13*rinv13*(rinvsq13-felec);
 463
 464             d                = r13-rswitch;
 465             d                = (d>0.0) ? d : 0.0;
 466             d2               = d*d;
 467             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 468
 469             dsw              = d2*(swF2+d*(swF3+d*swF4));
 470
 471             /* Evaluate switch function */
 472             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 473             felec            = felec*sw - rinv13*velec*dsw;
 474             velec           *= sw;
 475
 476             /* Update potential sums from outer loop */
 477             velecsum        += velec;
 478
 479             fscal            = felec;
 480
 481             /* Calculate temporary vectorial force */
 482             tx               = fscal*dx13;
 483             ty               = fscal*dy13;
 484             tz               = fscal*dz13;
 485
 486             /* Update vectorial force */
 487             fix1            += tx;
 488             fiy1            += ty;
 489             fiz1            += tz;
 490             f[j_coord_offset+DIM*3+XX] -= tx;
 491             f[j_coord_offset+DIM*3+YY] -= ty;
 492             f[j_coord_offset+DIM*3+ZZ] -= tz;
 493
 494             }
 495
 496             /**************************
 497              * CALCULATE INTERACTIONS *
 498              **************************/
 499
 500             if (rsq21<rcutoff2)
 501             {
 502
 503             r21              = rsq21*rinv21;
 504
 505             /* EWALD ELECTROSTATICS */
 506
 507             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 508             ewrt             = r21*ewtabscale;
 509             ewitab           = ewrt;
 510             eweps            = ewrt-ewitab;
 511             ewitab           = 4*ewitab;
 512             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 513             velec            = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 514             felec            = qq21*rinv21*(rinvsq21-felec);
 515
 516             d                = r21-rswitch;
 517             d                = (d>0.0) ? d : 0.0;
 518             d2               = d*d;
 519             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 520
 521             dsw              = d2*(swF2+d*(swF3+d*swF4));
 522
 523             /* Evaluate switch function */
 524             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 525             felec            = felec*sw - rinv21*velec*dsw;
 526             velec           *= sw;
 527
 528             /* Update potential sums from outer loop */
 529             velecsum        += velec;
 530
 531             fscal            = felec;
 532
 533             /* Calculate temporary vectorial force */
 534             tx               = fscal*dx21;
 535             ty               = fscal*dy21;
 536             tz               = fscal*dz21;
 537
 538             /* Update vectorial force */
 539             fix2            += tx;
 540             fiy2            += ty;
 541             fiz2            += tz;
 542             f[j_coord_offset+DIM*1+XX] -= tx;
 543             f[j_coord_offset+DIM*1+YY] -= ty;
 544             f[j_coord_offset+DIM*1+ZZ] -= tz;
 545
 546             }
 547
 548             /**************************
 549              * CALCULATE INTERACTIONS *
 550              **************************/
 551
 552             if (rsq22<rcutoff2)
 553             {
 554
 555             r22              = rsq22*rinv22;
 556
 557             /* EWALD ELECTROSTATICS */
 558
 559             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 560             ewrt             = r22*ewtabscale;
 561             ewitab           = ewrt;
 562             eweps            = ewrt-ewitab;
 563             ewitab           = 4*ewitab;
 564             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 565             velec            = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 566             felec            = qq22*rinv22*(rinvsq22-felec);
 567
 568             d                = r22-rswitch;
 569             d                = (d>0.0) ? d : 0.0;
 570             d2               = d*d;
 571             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 572
 573             dsw              = d2*(swF2+d*(swF3+d*swF4));
 574
 575             /* Evaluate switch function */
 576             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 577             felec            = felec*sw - rinv22*velec*dsw;
 578             velec           *= sw;
 579
 580             /* Update potential sums from outer loop */
 581             velecsum        += velec;
 582
 583             fscal            = felec;
 584
 585             /* Calculate temporary vectorial force */
 586             tx               = fscal*dx22;
 587             ty               = fscal*dy22;
 588             tz               = fscal*dz22;
 589
 590             /* Update vectorial force */
 591             fix2            += tx;
 592             fiy2            += ty;
 593             fiz2            += tz;
 594             f[j_coord_offset+DIM*2+XX] -= tx;
 595             f[j_coord_offset+DIM*2+YY] -= ty;
 596             f[j_coord_offset+DIM*2+ZZ] -= tz;
 597
 598             }
 599
 600             /**************************
 601              * CALCULATE INTERACTIONS *
 602              **************************/
 603
 604             if (rsq23<rcutoff2)
 605             {
 606
 607             r23              = rsq23*rinv23;
 608
 609             /* EWALD ELECTROSTATICS */
 610
 611             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 612             ewrt             = r23*ewtabscale;
 613             ewitab           = ewrt;
 614             eweps            = ewrt-ewitab;
 615             ewitab           = 4*ewitab;
 616             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 617             velec            = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 618             felec            = qq23*rinv23*(rinvsq23-felec);
 619
 620             d                = r23-rswitch;
 621             d                = (d>0.0) ? d : 0.0;
 622             d2               = d*d;
 623             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 624
 625             dsw              = d2*(swF2+d*(swF3+d*swF4));
 626
 627             /* Evaluate switch function */
 628             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 629             felec            = felec*sw - rinv23*velec*dsw;
 630             velec           *= sw;
 631
 632             /* Update potential sums from outer loop */
 633             velecsum        += velec;
 634
 635             fscal            = felec;
 636
 637             /* Calculate temporary vectorial force */
 638             tx               = fscal*dx23;
 639             ty               = fscal*dy23;
 640             tz               = fscal*dz23;
 641
 642             /* Update vectorial force */
 643             fix2            += tx;
 644             fiy2            += ty;
 645             fiz2            += tz;
 646             f[j_coord_offset+DIM*3+XX] -= tx;
 647             f[j_coord_offset+DIM*3+YY] -= ty;
 648             f[j_coord_offset+DIM*3+ZZ] -= tz;
 649
 650             }
 651
 652             /**************************
 653              * CALCULATE INTERACTIONS *
 654              **************************/
 655
 656             if (rsq31<rcutoff2)
 657             {
 658
 659             r31              = rsq31*rinv31;
 660
 661             /* EWALD ELECTROSTATICS */
 662
 663             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 664             ewrt             = r31*ewtabscale;
 665             ewitab           = ewrt;
 666             eweps            = ewrt-ewitab;
 667             ewitab           = 4*ewitab;
 668             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 669             velec            = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 670             felec            = qq31*rinv31*(rinvsq31-felec);
 671
 672             d                = r31-rswitch;
 673             d                = (d>0.0) ? d : 0.0;
 674             d2               = d*d;
 675             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 676
 677             dsw              = d2*(swF2+d*(swF3+d*swF4));
 678
 679             /* Evaluate switch function */
 680             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 681             felec            = felec*sw - rinv31*velec*dsw;
 682             velec           *= sw;
 683
 684             /* Update potential sums from outer loop */
 685             velecsum        += velec;
 686
 687             fscal            = felec;
 688
 689             /* Calculate temporary vectorial force */
 690             tx               = fscal*dx31;
 691             ty               = fscal*dy31;
 692             tz               = fscal*dz31;
 693
 694             /* Update vectorial force */
 695             fix3            += tx;
 696             fiy3            += ty;
 697             fiz3            += tz;
 698             f[j_coord_offset+DIM*1+XX] -= tx;
 699             f[j_coord_offset+DIM*1+YY] -= ty;
 700             f[j_coord_offset+DIM*1+ZZ] -= tz;
 701
 702             }
 703
 704             /**************************
 705              * CALCULATE INTERACTIONS *
 706              **************************/
 707
 708             if (rsq32<rcutoff2)
 709             {
 710
 711             r32              = rsq32*rinv32;
 712
 713             /* EWALD ELECTROSTATICS */
 714
 715             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 716             ewrt             = r32*ewtabscale;
 717             ewitab           = ewrt;
 718             eweps            = ewrt-ewitab;
 719             ewitab           = 4*ewitab;
 720             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 721             velec            = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 722             felec            = qq32*rinv32*(rinvsq32-felec);
 723
 724             d                = r32-rswitch;
 725             d                = (d>0.0) ? d : 0.0;
 726             d2               = d*d;
 727             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 728
 729             dsw              = d2*(swF2+d*(swF3+d*swF4));
 730
 731             /* Evaluate switch function */
 732             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 733             felec            = felec*sw - rinv32*velec*dsw;
 734             velec           *= sw;
 735
 736             /* Update potential sums from outer loop */
 737             velecsum        += velec;
 738
 739             fscal            = felec;
 740
 741             /* Calculate temporary vectorial force */
 742             tx               = fscal*dx32;
 743             ty               = fscal*dy32;
 744             tz               = fscal*dz32;
 745
 746             /* Update vectorial force */
 747             fix3            += tx;
 748             fiy3            += ty;
 749             fiz3            += tz;
 750             f[j_coord_offset+DIM*2+XX] -= tx;
 751             f[j_coord_offset+DIM*2+YY] -= ty;
 752             f[j_coord_offset+DIM*2+ZZ] -= tz;
 753
 754             }
 755
 756             /**************************
 757              * CALCULATE INTERACTIONS *
 758              **************************/
 759
 760             if (rsq33<rcutoff2)
 761             {
 762
 763             r33              = rsq33*rinv33;
 764
 765             /* EWALD ELECTROSTATICS */
 766
 767             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
 768             ewrt             = r33*ewtabscale;
 769             ewitab           = ewrt;
 770             eweps            = ewrt-ewitab;
 771             ewitab           = 4*ewitab;
 772             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
 773             velec            = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
 774             felec            = qq33*rinv33*(rinvsq33-felec);
 775
 776             d                = r33-rswitch;
 777             d                = (d>0.0) ? d : 0.0;
 778             d2               = d*d;
 779             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
 780
 781             dsw              = d2*(swF2+d*(swF3+d*swF4));
 782
 783             /* Evaluate switch function */
 784             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
 785             felec            = felec*sw - rinv33*velec*dsw;
 786             velec           *= sw;
 787
 788             /* Update potential sums from outer loop */
 789             velecsum        += velec;
 790
 791             fscal            = felec;
 792
 793             /* Calculate temporary vectorial force */
 794             tx               = fscal*dx33;
 795             ty               = fscal*dy33;
 796             tz               = fscal*dz33;
 797
 798             /* Update vectorial force */
 799             fix3            += tx;
 800             fiy3            += ty;
 801             fiz3            += tz;
 802             f[j_coord_offset+DIM*3+XX] -= tx;
 803             f[j_coord_offset+DIM*3+YY] -= ty;
 804             f[j_coord_offset+DIM*3+ZZ] -= tz;
 805
 806             }
 807
 808             /* Inner loop uses 601 flops */
 809         }
 810         /* End of innermost loop */
 811
 812         tx = ty = tz = 0;
 813         f[i_coord_offset+DIM*0+XX] += fix0;
 814         f[i_coord_offset+DIM*0+YY] += fiy0;
 815         f[i_coord_offset+DIM*0+ZZ] += fiz0;
 816         tx                         += fix0;
 817         ty                         += fiy0;
 818         tz                         += fiz0;
 819         f[i_coord_offset+DIM*1+XX] += fix1;
 820         f[i_coord_offset+DIM*1+YY] += fiy1;
 821         f[i_coord_offset+DIM*1+ZZ] += fiz1;
 822         tx                         += fix1;
 823         ty                         += fiy1;
 824         tz                         += fiz1;
 825         f[i_coord_offset+DIM*2+XX] += fix2;
 826         f[i_coord_offset+DIM*2+YY] += fiy2;
 827         f[i_coord_offset+DIM*2+ZZ] += fiz2;
 828         tx                         += fix2;
 829         ty                         += fiy2;
 830         tz                         += fiz2;
 831         f[i_coord_offset+DIM*3+XX] += fix3;
 832         f[i_coord_offset+DIM*3+YY] += fiy3;
 833         f[i_coord_offset+DIM*3+ZZ] += fiz3;
 834         tx                         += fix3;
 835         ty                         += fiy3;
 836         tz                         += fiz3;
 837         fshift[i_shift_offset+XX]  += tx;
 838         fshift[i_shift_offset+YY]  += ty;
 839         fshift[i_shift_offset+ZZ]  += tz;
 840
 841         ggid                        = gid[iidx];
 842         /* Update potential energies */
 843         kernel_data->energygrp_elec[ggid] += velecsum;
 844         kernel_data->energygrp_vdw[ggid] += vvdwsum;
 845
 846         /* Increment number of inner iterations */
 847         inneriter                  += j_index_end - j_index_start;
 848
 849         /* Outer loop uses 41 flops */
 850     }
 851
 852     /* Increment number of outer iterations */
 853     outeriter        += nri;
 854
 855     /* Update outer/inner flops */
 856
 857     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_VF,outeriter*41 + inneriter*601);
 858 }
 859 /*
 860  * Gromacs nonbonded kernel:   nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_F_c
 861  * Electrostatics interaction: Ewald
 862  * VdW interaction:            Buckingham
 863  * Geometry:                   Water4-Water4
 864  * Calculate force/pot:        Force
 865  */
 866 void
 867 nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4_F_c
 868                     (t_nblist * gmx_restrict                nlist,
 869                      rvec * gmx_restrict                    xx,
 870                      rvec * gmx_restrict                    ff,
 871                      t_forcerec * gmx_restrict              fr,
 872                      t_mdatoms * gmx_restrict               mdatoms,
 873                      nb_kernel_data_t * gmx_restrict        kernel_data,
 874                      t_nrnb * gmx_restrict                  nrnb)
 875 {
 876     int              i_shift_offset,i_coord_offset,j_coord_offset;
 877     int              j_index_start,j_index_end;
 878     int              nri,inr,ggid,iidx,jidx,jnr,outeriter,inneriter;
 879     real             shX,shY,shZ,tx,ty,tz,fscal,rcutoff,rcutoff2;
 880     int              *iinr,*jindex,*jjnr,*shiftidx,*gid;
 881     real             *shiftvec,*fshift,*x,*f;
 882     int              vdwioffset0;
 883     real             ix0,iy0,iz0,fix0,fiy0,fiz0,iq0,isai0;
 884     int              vdwioffset1;
 885     real             ix1,iy1,iz1,fix1,fiy1,fiz1,iq1,isai1;
 886     int              vdwioffset2;
 887     real             ix2,iy2,iz2,fix2,fiy2,fiz2,iq2,isai2;
 888     int              vdwioffset3;
 889     real             ix3,iy3,iz3,fix3,fiy3,fiz3,iq3,isai3;
 890     int              vdwjidx0;
 891     real             jx0,jy0,jz0,fjx0,fjy0,fjz0,jq0,isaj0;
 892     int              vdwjidx1;
 893     real             jx1,jy1,jz1,fjx1,fjy1,fjz1,jq1,isaj1;
 894     int              vdwjidx2;
 895     real             jx2,jy2,jz2,fjx2,fjy2,fjz2,jq2,isaj2;
 896     int              vdwjidx3;
 897     real             jx3,jy3,jz3,fjx3,fjy3,fjz3,jq3,isaj3;
 898     real             dx00,dy00,dz00,rsq00,rinv00,rinvsq00,r00,qq00,c6_00,c12_00,cexp1_00,cexp2_00;
 899     real             dx11,dy11,dz11,rsq11,rinv11,rinvsq11,r11,qq11,c6_11,c12_11,cexp1_11,cexp2_11;
 900     real             dx12,dy12,dz12,rsq12,rinv12,rinvsq12,r12,qq12,c6_12,c12_12,cexp1_12,cexp2_12;
 901     real             dx13,dy13,dz13,rsq13,rinv13,rinvsq13,r13,qq13,c6_13,c12_13,cexp1_13,cexp2_13;
 902     real             dx21,dy21,dz21,rsq21,rinv21,rinvsq21,r21,qq21,c6_21,c12_21,cexp1_21,cexp2_21;
 903     real             dx22,dy22,dz22,rsq22,rinv22,rinvsq22,r22,qq22,c6_22,c12_22,cexp1_22,cexp2_22;
 904     real             dx23,dy23,dz23,rsq23,rinv23,rinvsq23,r23,qq23,c6_23,c12_23,cexp1_23,cexp2_23;
 905     real             dx31,dy31,dz31,rsq31,rinv31,rinvsq31,r31,qq31,c6_31,c12_31,cexp1_31,cexp2_31;
 906     real             dx32,dy32,dz32,rsq32,rinv32,rinvsq32,r32,qq32,c6_32,c12_32,cexp1_32,cexp2_32;
 907     real             dx33,dy33,dz33,rsq33,rinv33,rinvsq33,r33,qq33,c6_33,c12_33,cexp1_33,cexp2_33;
 908     real             velec,felec,velecsum,facel,crf,krf,krf2;
 909     real             *charge;
 910     int              nvdwtype;
 911     real             rinvsix,rvdw,vvdw,vvdw6,vvdw12,fvdw,fvdw6,fvdw12,vvdwsum,br,vvdwexp,sh_vdw_invrcut6;
 912     int              *vdwtype;
 913     real             *vdwparam;
 914     int              ewitab;
 915     real             ewtabscale,eweps,sh_ewald,ewrt,ewtabhalfspace;
 916     real             *ewtab;
 917     real             rswitch,swV3,swV4,swV5,swF2,swF3,swF4,d,d2,sw,dsw;
 918
 919     x                = xx[0];
 920     f                = ff[0];
 921
 922     nri              = nlist->nri;
 923     iinr             = nlist->iinr;
 924     jindex           = nlist->jindex;
 925     jjnr             = nlist->jjnr;
 926     shiftidx         = nlist->shift;
 927     gid              = nlist->gid;
 928     shiftvec         = fr->shift_vec[0];
 929     fshift           = fr->fshift[0];
 930     facel            = fr->epsfac;
 931     charge           = mdatoms->chargeA;
 932     nvdwtype         = fr->ntype;
 933     vdwparam         = fr->nbfp;
 934     vdwtype          = mdatoms->typeA;
 935
 936     sh_ewald         = fr->ic->sh_ewald;
 937     ewtab            = fr->ic->tabq_coul_FDV0;
 938     ewtabscale       = fr->ic->tabq_scale;
 939     ewtabhalfspace   = 0.5/ewtabscale;
 940
 941     /* Setup water-specific parameters */
 942     inr              = nlist->iinr[0];
 943     iq1              = facel*charge[inr+1];
 944     iq2              = facel*charge[inr+2];
 945     iq3              = facel*charge[inr+3];
 946     vdwioffset0      = 3*nvdwtype*vdwtype[inr+0];
 947
 948     jq1              = charge[inr+1];
 949     jq2              = charge[inr+2];
 950     jq3              = charge[inr+3];
 951     vdwjidx0         = 3*vdwtype[inr+0];
 952     c6_00            = vdwparam[vdwioffset0+vdwjidx0];
 953     cexp1_00         = vdwparam[vdwioffset0+vdwjidx0+1];
 954     cexp2_00         = vdwparam[vdwioffset0+vdwjidx0+2];
 955     qq11             = iq1*jq1;
 956     qq12             = iq1*jq2;
 957     qq13             = iq1*jq3;
 958     qq21             = iq2*jq1;
 959     qq22             = iq2*jq2;
 960     qq23             = iq2*jq3;
 961     qq31             = iq3*jq1;
 962     qq32             = iq3*jq2;
 963     qq33             = iq3*jq3;
 964
 965     /* When we use explicit cutoffs the value must be identical for elec and VdW, so use elec as an arbitrary choice */
 966     rcutoff          = fr->rcoulomb;
 967     rcutoff2         = rcutoff*rcutoff;
 968
 969     rswitch          = fr->rcoulomb_switch;
 970     /* Setup switch parameters */
 971     d                = rcutoff-rswitch;
 972     swV3             = -10.0/(d*d*d);
 973     swV4             =  15.0/(d*d*d*d);
 974     swV5             =  -6.0/(d*d*d*d*d);
 975     swF2             = -30.0/(d*d*d);
 976     swF3             =  60.0/(d*d*d*d);
 977     swF4             = -30.0/(d*d*d*d*d);
 978
 979     outeriter        = 0;
 980     inneriter        = 0;
 981
 982     /* Start outer loop over neighborlists */
 983     for(iidx=0; iidx<nri; iidx++)
 984     {
 985         /* Load shift vector for this list */
 986         i_shift_offset   = DIM*shiftidx[iidx];
 987         shX              = shiftvec[i_shift_offset+XX];
 988         shY              = shiftvec[i_shift_offset+YY];
 989         shZ              = shiftvec[i_shift_offset+ZZ];
 990
 991         /* Load limits for loop over neighbors */
 992         j_index_start    = jindex[iidx];
 993         j_index_end      = jindex[iidx+1];
 994
 995         /* Get outer coordinate index */
 996         inr              = iinr[iidx];
 997         i_coord_offset   = DIM*inr;
 998
 999         /* Load i particle coords and add shift vector */
1000         ix0              = shX + x[i_coord_offset+DIM*0+XX];
1001         iy0              = shY + x[i_coord_offset+DIM*0+YY];
1002         iz0              = shZ + x[i_coord_offset+DIM*0+ZZ];
1003         ix1              = shX + x[i_coord_offset+DIM*1+XX];
1004         iy1              = shY + x[i_coord_offset+DIM*1+YY];
1005         iz1              = shZ + x[i_coord_offset+DIM*1+ZZ];
1006         ix2              = shX + x[i_coord_offset+DIM*2+XX];
1007         iy2              = shY + x[i_coord_offset+DIM*2+YY];
1008         iz2              = shZ + x[i_coord_offset+DIM*2+ZZ];
1009         ix3              = shX + x[i_coord_offset+DIM*3+XX];
1010         iy3              = shY + x[i_coord_offset+DIM*3+YY];
1011         iz3              = shZ + x[i_coord_offset+DIM*3+ZZ];
1012
1013         fix0             = 0.0;
1014         fiy0             = 0.0;
1015         fiz0             = 0.0;
1016         fix1             = 0.0;
1017         fiy1             = 0.0;
1018         fiz1             = 0.0;
1019         fix2             = 0.0;
1020         fiy2             = 0.0;
1021         fiz2             = 0.0;
1022         fix3             = 0.0;
1023         fiy3             = 0.0;
1024         fiz3             = 0.0;
1025
1026         /* Start inner kernel loop */
1027         for(jidx=j_index_start; jidx<j_index_end; jidx++)
1028         {
1029             /* Get j neighbor index, and coordinate index */
1030             jnr              = jjnr[jidx];
1031             j_coord_offset   = DIM*jnr;
1032
1033             /* load j atom coordinates */
1034             jx0              = x[j_coord_offset+DIM*0+XX];
1035             jy0              = x[j_coord_offset+DIM*0+YY];
1036             jz0              = x[j_coord_offset+DIM*0+ZZ];
1037             jx1              = x[j_coord_offset+DIM*1+XX];
1038             jy1              = x[j_coord_offset+DIM*1+YY];
1039             jz1              = x[j_coord_offset+DIM*1+ZZ];
1040             jx2              = x[j_coord_offset+DIM*2+XX];
1041             jy2              = x[j_coord_offset+DIM*2+YY];
1042             jz2              = x[j_coord_offset+DIM*2+ZZ];
1043             jx3              = x[j_coord_offset+DIM*3+XX];
1044             jy3              = x[j_coord_offset+DIM*3+YY];
1045             jz3              = x[j_coord_offset+DIM*3+ZZ];
1046
1047             /* Calculate displacement vector */
1048             dx00             = ix0 - jx0;
1049             dy00             = iy0 - jy0;
1050             dz00             = iz0 - jz0;
1051             dx11             = ix1 - jx1;
1052             dy11             = iy1 - jy1;
1053             dz11             = iz1 - jz1;
1054             dx12             = ix1 - jx2;
1055             dy12             = iy1 - jy2;
1056             dz12             = iz1 - jz2;
1057             dx13             = ix1 - jx3;
1058             dy13             = iy1 - jy3;
1059             dz13             = iz1 - jz3;
1060             dx21             = ix2 - jx1;
1061             dy21             = iy2 - jy1;
1062             dz21             = iz2 - jz1;
1063             dx22             = ix2 - jx2;
1064             dy22             = iy2 - jy2;
1065             dz22             = iz2 - jz2;
1066             dx23             = ix2 - jx3;
1067             dy23             = iy2 - jy3;
1068             dz23             = iz2 - jz3;
1069             dx31             = ix3 - jx1;
1070             dy31             = iy3 - jy1;
1071             dz31             = iz3 - jz1;
1072             dx32             = ix3 - jx2;
1073             dy32             = iy3 - jy2;
1074             dz32             = iz3 - jz2;
1075             dx33             = ix3 - jx3;
1076             dy33             = iy3 - jy3;
1077             dz33             = iz3 - jz3;
1078
1079             /* Calculate squared distance and things based on it */
1080             rsq00            = dx00*dx00+dy00*dy00+dz00*dz00;
1081             rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
1082             rsq12            = dx12*dx12+dy12*dy12+dz12*dz12;
1083             rsq13            = dx13*dx13+dy13*dy13+dz13*dz13;
1084             rsq21            = dx21*dx21+dy21*dy21+dz21*dz21;
1085             rsq22            = dx22*dx22+dy22*dy22+dz22*dz22;
1086             rsq23            = dx23*dx23+dy23*dy23+dz23*dz23;
1087             rsq31            = dx31*dx31+dy31*dy31+dz31*dz31;
1088             rsq32            = dx32*dx32+dy32*dy32+dz32*dz32;
1089             rsq33            = dx33*dx33+dy33*dy33+dz33*dz33;
1090
1091             rinv00           = gmx_invsqrt(rsq00);
1092             rinv11           = gmx_invsqrt(rsq11);
1093             rinv12           = gmx_invsqrt(rsq12);
1094             rinv13           = gmx_invsqrt(rsq13);
1095             rinv21           = gmx_invsqrt(rsq21);
1096             rinv22           = gmx_invsqrt(rsq22);
1097             rinv23           = gmx_invsqrt(rsq23);
1098             rinv31           = gmx_invsqrt(rsq31);
1099             rinv32           = gmx_invsqrt(rsq32);
1100             rinv33           = gmx_invsqrt(rsq33);
1101
1102             rinvsq00         = rinv00*rinv00;
1103             rinvsq11         = rinv11*rinv11;
1104             rinvsq12         = rinv12*rinv12;
1105             rinvsq13         = rinv13*rinv13;
1106             rinvsq21         = rinv21*rinv21;
1107             rinvsq22         = rinv22*rinv22;
1108             rinvsq23         = rinv23*rinv23;
1109             rinvsq31         = rinv31*rinv31;
1110             rinvsq32         = rinv32*rinv32;
1111             rinvsq33         = rinv33*rinv33;
1112
1113             /**************************
1114              * CALCULATE INTERACTIONS *
1115              **************************/
1116
1117             if (rsq00<rcutoff2)
1118             {
1119
1120             r00              = rsq00*rinv00;
1121
1122             /* BUCKINGHAM DISPERSION/REPULSION */
1123             rinvsix          = rinvsq00*rinvsq00*rinvsq00;
1124             vvdw6            = c6_00*rinvsix;
1125             br               = cexp2_00*r00;
1126             vvdwexp          = cexp1_00*exp(-br);
1127             vvdw             = vvdwexp - vvdw6*(1.0/6.0);
1128             fvdw             = (br*vvdwexp-vvdw6)*rinvsq00;
1129
1130             d                = r00-rswitch;
1131             d                = (d>0.0) ? d : 0.0;
1132             d2               = d*d;
1133             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1134
1135             dsw              = d2*(swF2+d*(swF3+d*swF4));
1136
1137             /* Evaluate switch function */
1138             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1139             fvdw             = fvdw*sw - rinv00*vvdw*dsw;
1140
1141             fscal            = fvdw;
1142
1143             /* Calculate temporary vectorial force */
1144             tx               = fscal*dx00;
1145             ty               = fscal*dy00;
1146             tz               = fscal*dz00;
1147
1148             /* Update vectorial force */
1149             fix0            += tx;
1150             fiy0            += ty;
1151             fiz0            += tz;
1152             f[j_coord_offset+DIM*0+XX] -= tx;
1153             f[j_coord_offset+DIM*0+YY] -= ty;
1154             f[j_coord_offset+DIM*0+ZZ] -= tz;
1155
1156             }
1157
1158             /**************************
1159              * CALCULATE INTERACTIONS *
1160              **************************/
1161
1162             if (rsq11<rcutoff2)
1163             {
1164
1165             r11              = rsq11*rinv11;
1166
1167             /* EWALD ELECTROSTATICS */
1168
1169             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1170             ewrt             = r11*ewtabscale;
1171             ewitab           = ewrt;
1172             eweps            = ewrt-ewitab;
1173             ewitab           = 4*ewitab;
1174             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1175             velec            = qq11*(rinv11-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1176             felec            = qq11*rinv11*(rinvsq11-felec);
1177
1178             d                = r11-rswitch;
1179             d                = (d>0.0) ? d : 0.0;
1180             d2               = d*d;
1181             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1182
1183             dsw              = d2*(swF2+d*(swF3+d*swF4));
1184
1185             /* Evaluate switch function */
1186             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1187             felec            = felec*sw - rinv11*velec*dsw;
1188
1189             fscal            = felec;
1190
1191             /* Calculate temporary vectorial force */
1192             tx               = fscal*dx11;
1193             ty               = fscal*dy11;
1194             tz               = fscal*dz11;
1195
1196             /* Update vectorial force */
1197             fix1            += tx;
1198             fiy1            += ty;
1199             fiz1            += tz;
1200             f[j_coord_offset+DIM*1+XX] -= tx;
1201             f[j_coord_offset+DIM*1+YY] -= ty;
1202             f[j_coord_offset+DIM*1+ZZ] -= tz;
1203
1204             }
1205
1206             /**************************
1207              * CALCULATE INTERACTIONS *
1208              **************************/
1209
1210             if (rsq12<rcutoff2)
1211             {
1212
1213             r12              = rsq12*rinv12;
1214
1215             /* EWALD ELECTROSTATICS */
1216
1217             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1218             ewrt             = r12*ewtabscale;
1219             ewitab           = ewrt;
1220             eweps            = ewrt-ewitab;
1221             ewitab           = 4*ewitab;
1222             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1223             velec            = qq12*(rinv12-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1224             felec            = qq12*rinv12*(rinvsq12-felec);
1225
1226             d                = r12-rswitch;
1227             d                = (d>0.0) ? d : 0.0;
1228             d2               = d*d;
1229             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1230
1231             dsw              = d2*(swF2+d*(swF3+d*swF4));
1232
1233             /* Evaluate switch function */
1234             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1235             felec            = felec*sw - rinv12*velec*dsw;
1236
1237             fscal            = felec;
1238
1239             /* Calculate temporary vectorial force */
1240             tx               = fscal*dx12;
1241             ty               = fscal*dy12;
1242             tz               = fscal*dz12;
1243
1244             /* Update vectorial force */
1245             fix1            += tx;
1246             fiy1            += ty;
1247             fiz1            += tz;
1248             f[j_coord_offset+DIM*2+XX] -= tx;
1249             f[j_coord_offset+DIM*2+YY] -= ty;
1250             f[j_coord_offset+DIM*2+ZZ] -= tz;
1251
1252             }
1253
1254             /**************************
1255              * CALCULATE INTERACTIONS *
1256              **************************/
1257
1258             if (rsq13<rcutoff2)
1259             {
1260
1261             r13              = rsq13*rinv13;
1262
1263             /* EWALD ELECTROSTATICS */
1264
1265             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1266             ewrt             = r13*ewtabscale;
1267             ewitab           = ewrt;
1268             eweps            = ewrt-ewitab;
1269             ewitab           = 4*ewitab;
1270             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1271             velec            = qq13*(rinv13-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1272             felec            = qq13*rinv13*(rinvsq13-felec);
1273
1274             d                = r13-rswitch;
1275             d                = (d>0.0) ? d : 0.0;
1276             d2               = d*d;
1277             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1278
1279             dsw              = d2*(swF2+d*(swF3+d*swF4));
1280
1281             /* Evaluate switch function */
1282             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1283             felec            = felec*sw - rinv13*velec*dsw;
1284
1285             fscal            = felec;
1286
1287             /* Calculate temporary vectorial force */
1288             tx               = fscal*dx13;
1289             ty               = fscal*dy13;
1290             tz               = fscal*dz13;
1291
1292             /* Update vectorial force */
1293             fix1            += tx;
1294             fiy1            += ty;
1295             fiz1            += tz;
1296             f[j_coord_offset+DIM*3+XX] -= tx;
1297             f[j_coord_offset+DIM*3+YY] -= ty;
1298             f[j_coord_offset+DIM*3+ZZ] -= tz;
1299
1300             }
1301
1302             /**************************
1303              * CALCULATE INTERACTIONS *
1304              **************************/
1305
1306             if (rsq21<rcutoff2)
1307             {
1308
1309             r21              = rsq21*rinv21;
1310
1311             /* EWALD ELECTROSTATICS */
1312
1313             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1314             ewrt             = r21*ewtabscale;
1315             ewitab           = ewrt;
1316             eweps            = ewrt-ewitab;
1317             ewitab           = 4*ewitab;
1318             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1319             velec            = qq21*(rinv21-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1320             felec            = qq21*rinv21*(rinvsq21-felec);
1321
1322             d                = r21-rswitch;
1323             d                = (d>0.0) ? d : 0.0;
1324             d2               = d*d;
1325             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1326
1327             dsw              = d2*(swF2+d*(swF3+d*swF4));
1328
1329             /* Evaluate switch function */
1330             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1331             felec            = felec*sw - rinv21*velec*dsw;
1332
1333             fscal            = felec;
1334
1335             /* Calculate temporary vectorial force */
1336             tx               = fscal*dx21;
1337             ty               = fscal*dy21;
1338             tz               = fscal*dz21;
1339
1340             /* Update vectorial force */
1341             fix2            += tx;
1342             fiy2            += ty;
1343             fiz2            += tz;
1344             f[j_coord_offset+DIM*1+XX] -= tx;
1345             f[j_coord_offset+DIM*1+YY] -= ty;
1346             f[j_coord_offset+DIM*1+ZZ] -= tz;
1347
1348             }
1349
1350             /**************************
1351              * CALCULATE INTERACTIONS *
1352              **************************/
1353
1354             if (rsq22<rcutoff2)
1355             {
1356
1357             r22              = rsq22*rinv22;
1358
1359             /* EWALD ELECTROSTATICS */
1360
1361             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1362             ewrt             = r22*ewtabscale;
1363             ewitab           = ewrt;
1364             eweps            = ewrt-ewitab;
1365             ewitab           = 4*ewitab;
1366             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1367             velec            = qq22*(rinv22-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1368             felec            = qq22*rinv22*(rinvsq22-felec);
1369
1370             d                = r22-rswitch;
1371             d                = (d>0.0) ? d : 0.0;
1372             d2               = d*d;
1373             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1374
1375             dsw              = d2*(swF2+d*(swF3+d*swF4));
1376
1377             /* Evaluate switch function */
1378             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1379             felec            = felec*sw - rinv22*velec*dsw;
1380
1381             fscal            = felec;
1382
1383             /* Calculate temporary vectorial force */
1384             tx               = fscal*dx22;
1385             ty               = fscal*dy22;
1386             tz               = fscal*dz22;
1387
1388             /* Update vectorial force */
1389             fix2            += tx;
1390             fiy2            += ty;
1391             fiz2            += tz;
1392             f[j_coord_offset+DIM*2+XX] -= tx;
1393             f[j_coord_offset+DIM*2+YY] -= ty;
1394             f[j_coord_offset+DIM*2+ZZ] -= tz;
1395
1396             }
1397
1398             /**************************
1399              * CALCULATE INTERACTIONS *
1400              **************************/
1401
1402             if (rsq23<rcutoff2)
1403             {
1404
1405             r23              = rsq23*rinv23;
1406
1407             /* EWALD ELECTROSTATICS */
1408
1409             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1410             ewrt             = r23*ewtabscale;
1411             ewitab           = ewrt;
1412             eweps            = ewrt-ewitab;
1413             ewitab           = 4*ewitab;
1414             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1415             velec            = qq23*(rinv23-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1416             felec            = qq23*rinv23*(rinvsq23-felec);
1417
1418             d                = r23-rswitch;
1419             d                = (d>0.0) ? d : 0.0;
1420             d2               = d*d;
1421             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1422
1423             dsw              = d2*(swF2+d*(swF3+d*swF4));
1424
1425             /* Evaluate switch function */
1426             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1427             felec            = felec*sw - rinv23*velec*dsw;
1428
1429             fscal            = felec;
1430
1431             /* Calculate temporary vectorial force */
1432             tx               = fscal*dx23;
1433             ty               = fscal*dy23;
1434             tz               = fscal*dz23;
1435
1436             /* Update vectorial force */
1437             fix2            += tx;
1438             fiy2            += ty;
1439             fiz2            += tz;
1440             f[j_coord_offset+DIM*3+XX] -= tx;
1441             f[j_coord_offset+DIM*3+YY] -= ty;
1442             f[j_coord_offset+DIM*3+ZZ] -= tz;
1443
1444             }
1445
1446             /**************************
1447              * CALCULATE INTERACTIONS *
1448              **************************/
1449
1450             if (rsq31<rcutoff2)
1451             {
1452
1453             r31              = rsq31*rinv31;
1454
1455             /* EWALD ELECTROSTATICS */
1456
1457             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1458             ewrt             = r31*ewtabscale;
1459             ewitab           = ewrt;
1460             eweps            = ewrt-ewitab;
1461             ewitab           = 4*ewitab;
1462             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1463             velec            = qq31*(rinv31-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1464             felec            = qq31*rinv31*(rinvsq31-felec);
1465
1466             d                = r31-rswitch;
1467             d                = (d>0.0) ? d : 0.0;
1468             d2               = d*d;
1469             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1470
1471             dsw              = d2*(swF2+d*(swF3+d*swF4));
1472
1473             /* Evaluate switch function */
1474             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1475             felec            = felec*sw - rinv31*velec*dsw;
1476
1477             fscal            = felec;
1478
1479             /* Calculate temporary vectorial force */
1480             tx               = fscal*dx31;
1481             ty               = fscal*dy31;
1482             tz               = fscal*dz31;
1483
1484             /* Update vectorial force */
1485             fix3            += tx;
1486             fiy3            += ty;
1487             fiz3            += tz;
1488             f[j_coord_offset+DIM*1+XX] -= tx;
1489             f[j_coord_offset+DIM*1+YY] -= ty;
1490             f[j_coord_offset+DIM*1+ZZ] -= tz;
1491
1492             }
1493
1494             /**************************
1495              * CALCULATE INTERACTIONS *
1496              **************************/
1497
1498             if (rsq32<rcutoff2)
1499             {
1500
1501             r32              = rsq32*rinv32;
1502
1503             /* EWALD ELECTROSTATICS */
1504
1505             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1506             ewrt             = r32*ewtabscale;
1507             ewitab           = ewrt;
1508             eweps            = ewrt-ewitab;
1509             ewitab           = 4*ewitab;
1510             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1511             velec            = qq32*(rinv32-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1512             felec            = qq32*rinv32*(rinvsq32-felec);
1513
1514             d                = r32-rswitch;
1515             d                = (d>0.0) ? d : 0.0;
1516             d2               = d*d;
1517             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1518
1519             dsw              = d2*(swF2+d*(swF3+d*swF4));
1520
1521             /* Evaluate switch function */
1522             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1523             felec            = felec*sw - rinv32*velec*dsw;
1524
1525             fscal            = felec;
1526
1527             /* Calculate temporary vectorial force */
1528             tx               = fscal*dx32;
1529             ty               = fscal*dy32;
1530             tz               = fscal*dz32;
1531
1532             /* Update vectorial force */
1533             fix3            += tx;
1534             fiy3            += ty;
1535             fiz3            += tz;
1536             f[j_coord_offset+DIM*2+XX] -= tx;
1537             f[j_coord_offset+DIM*2+YY] -= ty;
1538             f[j_coord_offset+DIM*2+ZZ] -= tz;
1539
1540             }
1541
1542             /**************************
1543              * CALCULATE INTERACTIONS *
1544              **************************/
1545
1546             if (rsq33<rcutoff2)
1547             {
1548
1549             r33              = rsq33*rinv33;
1550
1551             /* EWALD ELECTROSTATICS */
1552
1553             /* Calculate Ewald table index by multiplying r with scale and truncate to integer */
1554             ewrt             = r33*ewtabscale;
1555             ewitab           = ewrt;
1556             eweps            = ewrt-ewitab;
1557             ewitab           = 4*ewitab;
1558             felec            = ewtab[ewitab]+eweps*ewtab[ewitab+1];
1559             velec            = qq33*(rinv33-(ewtab[ewitab+2]-ewtabhalfspace*eweps*(ewtab[ewitab]+felec)));
1560             felec            = qq33*rinv33*(rinvsq33-felec);
1561
1562             d                = r33-rswitch;
1563             d                = (d>0.0) ? d : 0.0;
1564             d2               = d*d;
1565             sw               = 1.0+d2*d*(swV3+d*(swV4+d*swV5));
1566
1567             dsw              = d2*(swF2+d*(swF3+d*swF4));
1568
1569             /* Evaluate switch function */
1570             /* fscal'=f'/r=-(v*sw)'/r=-(v'*sw+v*dsw)/r=-v'*sw/r-v*dsw/r=fscal*sw-v*dsw/r */
1571             felec            = felec*sw - rinv33*velec*dsw;
1572
1573             fscal            = felec;
1574
1575             /* Calculate temporary vectorial force */
1576             tx               = fscal*dx33;
1577             ty               = fscal*dy33;
1578             tz               = fscal*dz33;
1579
1580             /* Update vectorial force */
1581             fix3            += tx;
1582             fiy3            += ty;
1583             fiz3            += tz;
1584             f[j_coord_offset+DIM*3+XX] -= tx;
1585             f[j_coord_offset+DIM*3+YY] -= ty;
1586             f[j_coord_offset+DIM*3+ZZ] -= tz;
1587
1588             }
1589
1590             /* Inner loop uses 581 flops */
1591         }
1592         /* End of innermost loop */
1593
1594         tx = ty = tz = 0;
1595         f[i_coord_offset+DIM*0+XX] += fix0;
1596         f[i_coord_offset+DIM*0+YY] += fiy0;
1597         f[i_coord_offset+DIM*0+ZZ] += fiz0;
1598         tx                         += fix0;
1599         ty                         += fiy0;
1600         tz                         += fiz0;
1601         f[i_coord_offset+DIM*1+XX] += fix1;
1602         f[i_coord_offset+DIM*1+YY] += fiy1;
1603         f[i_coord_offset+DIM*1+ZZ] += fiz1;
1604         tx                         += fix1;
1605         ty                         += fiy1;
1606         tz                         += fiz1;
1607         f[i_coord_offset+DIM*2+XX] += fix2;
1608         f[i_coord_offset+DIM*2+YY] += fiy2;
1609         f[i_coord_offset+DIM*2+ZZ] += fiz2;
1610         tx                         += fix2;
1611         ty                         += fiy2;
1612         tz                         += fiz2;
1613         f[i_coord_offset+DIM*3+XX] += fix3;
1614         f[i_coord_offset+DIM*3+YY] += fiy3;
1615         f[i_coord_offset+DIM*3+ZZ] += fiz3;
1616         tx                         += fix3;
1617         ty                         += fiy3;
1618         tz                         += fiz3;
1619         fshift[i_shift_offset+XX]  += tx;
1620         fshift[i_shift_offset+YY]  += ty;
1621         fshift[i_shift_offset+ZZ]  += tz;
1622
1623         /* Increment number of inner iterations */
1624         inneriter                  += j_index_end - j_index_start;
1625
1626         /* Outer loop uses 39 flops */
1627     }
1628
1629     /* Increment number of outer iterations */
1630     outeriter        += nri;
1631
1632     /* Update outer/inner flops */
1633
1634     inc_nrnb(nrnb,eNR_NBKERNEL_ELEC_VDW_W4W4_F,outeriter*39 + inneriter*581);
1635 }