src/gromacs/gmxlib/nonbonded/nb_kernel_adress_c/nb_kernel310_c_adress.c

   1 /*
   2  * Copyright (c) Erik Lindahl, David van der Spoel 2003
   3  *
   4  * This file is generated automatically at compile time
   5  * by the program mknb in the Gromacs distribution.
   6  *
   7  * Options used when generation this file:
   8  * Language:         c
   9  * Precision:        single
  10  * Threads:          yes
  11  * Software invsqrt: no
  12  * PowerPC invsqrt:  no
  13  * Prefetch forces:  no
  14  * Adress kernel:  yes
  15  * Comments:         no
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include<config.h>
  19 #endif
  20 #ifdef GMX_THREAD_SHM_FDECOMP
  21 #include<thread_mpi.h>
  22 #endif
  23 #define ALMOST_ZERO 1e-30
  24 #define ALMOST_ONE 1-(1e-30)
  25 #include<math.h>
  26
  27 #include "nb_kernel310_adress.h"
  28
  29
  30
  31 /*
  32  * Gromacs nonbonded kernel nb_kernel310_adress_cg
  33  * Coulomb interaction:     Tabulated
  34  * VdW interaction:         Lennard-Jones
  35  * water optimization:      No
  36  * Calculate forces:        yes
  37  */
  38 void nb_kernel310_adress_cg(
  39                     int *           p_nri,
  40                     int *           iinr,
  41                     int *           jindex,
  42                     int *           jjnr,
  43                     int *           shift,
  44                     real *         shiftvec,
  45                     real *         fshift,
  46                     int *           gid,
  47                     real *         pos,
  48                     real *         faction,
  49                     real *         charge,
  50                     real *         p_facel,
  51                     real *         p_krf,
  52                     real *         p_crf,
  53                     real *         Vc,
  54                     int *           type,
  55                     int *           p_ntype,
  56                     real *         vdwparam,
  57                     real *         Vvdw,
  58                     real *         p_tabscale,
  59                     real *         VFtab,
  60                     real *         invsqrta,
  61                     real *         dvda,
  62                     real *         p_gbtabscale,
  63                     real *         GBtab,
  64                     int *           p_nthreads,
  65                     int *           count,
  66                     void *          mtx,
  67                     int *           outeriter,
  68                     int *           inneriter,
  69                     real           force_cap,
  70                     real *         wf)
  71 {
  72     int           nri,ntype,nthreads;
  73     real         facel,krf,crf,tabscale,gbtabscale;
  74     int           n,ii,is3,ii3,k,nj0,nj1,jnr,j3,ggid;
  75     int           nn0,nn1,nouter,ninner;
  76     real         shX,shY,shZ;
  77     real         fscal,tx,ty,tz;
  78     real         rinvsq;
  79     real         iq;
  80     real         qq,vcoul,vctot;
  81     int           nti;
  82     int           tj;
  83     real         rinvsix;
  84     real         Vvdw6,Vvdwtot;
  85     real         Vvdw12;
  86     real         r,rt,eps,eps2;
  87     int           n0,nnn;
  88     real         Y,F,Geps,Heps2,Fp,VV;
  89     real         FF;
  90     real         fijC;
  91     real         ix1,iy1,iz1,fix1,fiy1,fiz1;
  92     real         jx1,jy1,jz1;
  93     real         dx11,dy11,dz11,rsq11,rinv11;
  94     real         c6,c12;
  95     real         weight_cg1, weight_cg2, weight_product;
  96     real         hybscal;
  97
  98     nri              = *p_nri;
  99     ntype            = *p_ntype;
 100     nthreads         = *p_nthreads;
 101     facel            = *p_facel;
 102     krf              = *p_krf;
 103     crf              = *p_crf;
 104     tabscale         = *p_tabscale;
 105     nouter           = 0;
 106     ninner           = 0;
 107
 108     do
 109     {
 110         #ifdef GMX_THREAD_SHM_FDECOMP
 111         tMPI_Thread_mutex_lock((tMPI_Thread_mutex_t *)mtx);
 112         nn0              = *count;
 113         nn1              = nn0+(nri-nn0)/(2*nthreads)+10;
 114         *count           = nn1;
 115         tMPI_Thread_mutex_unlock((tMPI_Thread_mutex_t *)mtx);
 116         if(nn1>nri) nn1=nri;
 117         #else
 118         nn0 = 0;
 119         nn1 = nri;
 120         #endif
 121
 122         for(n=nn0; (n<nn1); n++)
 123         {
 124             is3              = 3*shift[n];
 125             shX              = shiftvec[is3];
 126             shY              = shiftvec[is3+1];
 127             shZ              = shiftvec[is3+2];
 128             nj0              = jindex[n];
 129             nj1              = jindex[n+1];
 130             ii               = iinr[n];
 131             ii3              = 3*ii;
 132             ix1              = shX + pos[ii3+0];
 133             iy1              = shY + pos[ii3+1];
 134             iz1              = shZ + pos[ii3+2];
 135             iq               = facel*charge[ii];
 136             nti              = 2*ntype*type[ii];
 137             weight_cg1       = wf[ii];
 138             vctot            = 0;
 139             Vvdwtot          = 0;
 140             fix1             = 0;
 141             fiy1             = 0;
 142             fiz1             = 0;
 143
 144             for(k=nj0; (k<nj1); k++)
 145             {
 146                 jnr              = jjnr[k];
 147                 weight_cg2       = wf[jnr];
 148                 weight_product   = weight_cg1*weight_cg2;
 149                 if (weight_product < ALMOST_ZERO) {
 150                        hybscal = 1.0;
 151                 }
 152                 else if (weight_product >= ALMOST_ONE)
 153                 {
 154                   /* force is zero, skip this molecule */
 155                        continue;
 156                 }
 157                 else
 158                 {
 159                    hybscal = 1.0 - weight_product;
 160                 }
 161                 j3               = 3*jnr;
 162                 jx1              = pos[j3+0];
 163                 jy1              = pos[j3+1];
 164                 jz1              = pos[j3+2];
 165                 dx11             = ix1 - jx1;
 166                 dy11             = iy1 - jy1;
 167                 dz11             = iz1 - jz1;
 168                 rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 169                 rinv11           = 1.0/sqrt(rsq11);
 170                 qq               = iq*charge[jnr];
 171                 tj               = nti+2*type[jnr];
 172                 c6               = vdwparam[tj];
 173                 c12              = vdwparam[tj+1];
 174                 rinvsq           = rinv11*rinv11;
 175                 r                = rsq11*rinv11;
 176                 rt               = r*tabscale;
 177                 n0               = rt;
 178                 eps              = rt-n0;
 179                 eps2             = eps*eps;
 180                 nnn              = 4*n0;
 181                 Y                = VFtab[nnn];
 182                 F                = VFtab[nnn+1];
 183                 Geps             = eps*VFtab[nnn+2];
 184                 Heps2            = eps2*VFtab[nnn+3];
 185                 Fp               = F+Geps+Heps2;
 186                 VV               = Y+eps*Fp;
 187                 FF               = Fp+Geps+2.0*Heps2;
 188                 vcoul            = qq*VV;
 189                 fijC             = qq*FF;
 190                 vctot            = vctot + vcoul;
 191                 rinvsix          = rinvsq*rinvsq*rinvsq;
 192                 Vvdw6            = c6*rinvsix;
 193                 Vvdw12           = c12*rinvsix*rinvsix;
 194                 Vvdwtot          = Vvdwtot+Vvdw12-Vvdw6;
 195                 fscal            = (12.0*Vvdw12-6.0*Vvdw6)*rinvsq-((fijC)*tabscale)*rinv11;
 196                 fscal *= hybscal;
 197                 tx               = fscal*dx11;
 198                 ty               = fscal*dy11;
 199                 tz               = fscal*dz11;
 200                 fix1             = fix1 + tx;
 201                 fiy1             = fiy1 + ty;
 202                 fiz1             = fiz1 + tz;
 203                 faction[j3+0]    = faction[j3+0] - tx;
 204                 faction[j3+1]    = faction[j3+1] - ty;
 205                 faction[j3+2]    = faction[j3+2] - tz;
 206             }
 207
 208             faction[ii3+0]   = faction[ii3+0] + fix1;
 209             faction[ii3+1]   = faction[ii3+1] + fiy1;
 210             faction[ii3+2]   = faction[ii3+2] + fiz1;
 211             fshift[is3]      = fshift[is3]+fix1;
 212             fshift[is3+1]    = fshift[is3+1]+fiy1;
 213             fshift[is3+2]    = fshift[is3+2]+fiz1;
 214             ggid             = gid[n];
 215             Vc[ggid]         = Vc[ggid] + vctot;
 216             Vvdw[ggid]       = Vvdw[ggid] + Vvdwtot;
 217             ninner           = ninner + nj1 - nj0;
 218         }
 219
 220         nouter           = nouter + nn1 - nn0;
 221     }
 222     while (nn1<nri);
 223
 224     *outeriter       = nouter;
 225     *inneriter       = ninner;
 226 }
 227
 228
 229
 230
 231
 232 /*
 233  * Gromacs nonbonded kernel nb_kernel310_adress_ex
 234  * Coulomb interaction:     Tabulated
 235  * VdW interaction:         Lennard-Jones
 236  * water optimization:      No
 237  * Calculate forces:        yes
 238  */
 239 void nb_kernel310_adress_ex(
 240                     int *           p_nri,
 241                     int *           iinr,
 242                     int *           jindex,
 243                     int *           jjnr,
 244                     int *           shift,
 245                     real *         shiftvec,
 246                     real *         fshift,
 247                     int *           gid,
 248                     real *         pos,
 249                     real *         faction,
 250                     real *         charge,
 251                     real *         p_facel,
 252                     real *         p_krf,
 253                     real *         p_crf,
 254                     real *         Vc,
 255                     int *           type,
 256                     int *           p_ntype,
 257                     real *         vdwparam,
 258                     real *         Vvdw,
 259                     real *         p_tabscale,
 260                     real *         VFtab,
 261                     real *         invsqrta,
 262                     real *         dvda,
 263                     real *         p_gbtabscale,
 264                     real *         GBtab,
 265                     int *           p_nthreads,
 266                     int *           count,
 267                     void *          mtx,
 268                     int *           outeriter,
 269                     int *           inneriter,
 270                     real           force_cap,
 271                     real *         wf)
 272 {
 273     int           nri,ntype,nthreads;
 274     real         facel,krf,crf,tabscale,gbtabscale;
 275     int           n,ii,is3,ii3,k,nj0,nj1,jnr,j3,ggid;
 276     int           nn0,nn1,nouter,ninner;
 277     real         shX,shY,shZ;
 278     real         fscal,tx,ty,tz;
 279     real         rinvsq;
 280     real         iq;
 281     real         qq,vcoul,vctot;
 282     int           nti;
 283     int           tj;
 284     real         rinvsix;
 285     real         Vvdw6,Vvdwtot;
 286     real         Vvdw12;
 287     real         r,rt,eps,eps2;
 288     int           n0,nnn;
 289     real         Y,F,Geps,Heps2,Fp,VV;
 290     real         FF;
 291     real         fijC;
 292     real         ix1,iy1,iz1,fix1,fiy1,fiz1;
 293     real         jx1,jy1,jz1;
 294     real         dx11,dy11,dz11,rsq11,rinv11;
 295     real         c6,c12;
 296     real         weight_cg1, weight_cg2, weight_product;
 297     real         hybscal;
 298
 299     nri              = *p_nri;
 300     ntype            = *p_ntype;
 301     nthreads         = *p_nthreads;
 302     facel            = *p_facel;
 303     krf              = *p_krf;
 304     crf              = *p_crf;
 305     tabscale         = *p_tabscale;
 306     nouter           = 0;
 307     ninner           = 0;
 308
 309     do
 310     {
 311         #ifdef GMX_THREAD_SHM_FDECOMP
 312         tMPI_Thread_mutex_lock((tMPI_Thread_mutex_t *)mtx);
 313         nn0              = *count;
 314         nn1              = nn0+(nri-nn0)/(2*nthreads)+10;
 315         *count           = nn1;
 316         tMPI_Thread_mutex_unlock((tMPI_Thread_mutex_t *)mtx);
 317         if(nn1>nri) nn1=nri;
 318         #else
 319         nn0 = 0;
 320         nn1 = nri;
 321         #endif
 322
 323         for(n=nn0; (n<nn1); n++)
 324         {
 325             is3              = 3*shift[n];
 326             shX              = shiftvec[is3];
 327             shY              = shiftvec[is3+1];
 328             shZ              = shiftvec[is3+2];
 329             nj0              = jindex[n];
 330             nj1              = jindex[n+1];
 331             ii               = iinr[n];
 332             ii3              = 3*ii;
 333             ix1              = shX + pos[ii3+0];
 334             iy1              = shY + pos[ii3+1];
 335             iz1              = shZ + pos[ii3+2];
 336             iq               = facel*charge[ii];
 337             nti              = 2*ntype*type[ii];
 338             weight_cg1       = wf[ii];
 339             vctot            = 0;
 340             Vvdwtot          = 0;
 341             fix1             = 0;
 342             fiy1             = 0;
 343             fiz1             = 0;
 344
 345             for(k=nj0; (k<nj1); k++)
 346             {
 347                 jnr              = jjnr[k];
 348                 weight_cg2       = wf[jnr];
 349                 weight_product   = weight_cg1*weight_cg2;
 350                 if (weight_product < ALMOST_ZERO) {
 351                 /* force is zero, skip this molecule */
 352                  continue;
 353                 }
 354                 else if (weight_product >= ALMOST_ONE)
 355                 {
 356                        hybscal = 1.0;
 357                 }
 358                 else
 359                 {
 360                    hybscal = weight_product;
 361                 }
 362                 j3               = 3*jnr;
 363                 jx1              = pos[j3+0];
 364                 jy1              = pos[j3+1];
 365                 jz1              = pos[j3+2];
 366                 dx11             = ix1 - jx1;
 367                 dy11             = iy1 - jy1;
 368                 dz11             = iz1 - jz1;
 369                 rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 370                 rinv11           = 1.0/sqrt(rsq11);
 371                 qq               = iq*charge[jnr];
 372                 tj               = nti+2*type[jnr];
 373                 c6               = vdwparam[tj];
 374                 c12              = vdwparam[tj+1];
 375                 rinvsq           = rinv11*rinv11;
 376                 r                = rsq11*rinv11;
 377                 rt               = r*tabscale;
 378                 n0               = rt;
 379                 eps              = rt-n0;
 380                 eps2             = eps*eps;
 381                 nnn              = 4*n0;
 382                 Y                = VFtab[nnn];
 383                 F                = VFtab[nnn+1];
 384                 Geps             = eps*VFtab[nnn+2];
 385                 Heps2            = eps2*VFtab[nnn+3];
 386                 Fp               = F+Geps+Heps2;
 387                 VV               = Y+eps*Fp;
 388                 FF               = Fp+Geps+2.0*Heps2;
 389                 vcoul            = qq*VV;
 390                 fijC             = qq*FF;
 391                 vctot            = vctot + vcoul;
 392                 rinvsix          = rinvsq*rinvsq*rinvsq;
 393                 Vvdw6            = c6*rinvsix;
 394                 Vvdw12           = c12*rinvsix*rinvsix;
 395                 Vvdwtot          = Vvdwtot+Vvdw12-Vvdw6;
 396                 fscal            = (12.0*Vvdw12-6.0*Vvdw6)*rinvsq-((fijC)*tabscale)*rinv11;
 397                 fscal *= hybscal;
 398                 if(force_cap>0 && (fabs(fscal)> force_cap)){
 399                 fscal=force_cap*fscal/fabs(fscal);
 400                 }
 401                 tx               = fscal*dx11;
 402                 ty               = fscal*dy11;
 403                 tz               = fscal*dz11;
 404                 fix1             = fix1 + tx;
 405                 fiy1             = fiy1 + ty;
 406                 fiz1             = fiz1 + tz;
 407                 faction[j3+0]    = faction[j3+0] - tx;
 408                 faction[j3+1]    = faction[j3+1] - ty;
 409                 faction[j3+2]    = faction[j3+2] - tz;
 410             }
 411
 412             faction[ii3+0]   = faction[ii3+0] + fix1;
 413             faction[ii3+1]   = faction[ii3+1] + fiy1;
 414             faction[ii3+2]   = faction[ii3+2] + fiz1;
 415             fshift[is3]      = fshift[is3]+fix1;
 416             fshift[is3+1]    = fshift[is3+1]+fiy1;
 417             fshift[is3+2]    = fshift[is3+2]+fiz1;
 418             ggid             = gid[n];
 419             Vc[ggid]         = Vc[ggid] + vctot;
 420             Vvdw[ggid]       = Vvdw[ggid] + Vvdwtot;
 421             ninner           = ninner + nj1 - nj0;
 422         }
 423
 424         nouter           = nouter + nn1 - nn0;
 425     }
 426     while (nn1<nri);
 427
 428     *outeriter       = nouter;
 429     *inneriter       = ninner;
 430 }
 431
 432