src/gromacs/gmxlib/nonbonded/nb_kernel_adress_c/nb_kernel320_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_kernel320_adress.h"
  28
  29
  30
  31 /*
  32  * Gromacs nonbonded kernel nb_kernel320_adress_cg
  33  * Coulomb interaction:     Tabulated
  34  * VdW interaction:         Buckingham
  35  * water optimization:      No
  36  * Calculate forces:        yes
  37  */
  38 void nb_kernel320_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         r,rt,eps,eps2;
  86     int           n0,nnn;
  87     real         Y,F,Geps,Heps2,Fp,VV;
  88     real         FF;
  89     real         fijC;
  90     real         Vvdwexp,br;
  91     real         ix1,iy1,iz1,fix1,fiy1,fiz1;
  92     real         jx1,jy1,jz1;
  93     real         dx11,dy11,dz11,rsq11,rinv11;
  94     real         c6,cexp1,cexp2;
  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              = 3*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+3*type[jnr];
 172                 c6               = vdwparam[tj];
 173                 cexp1            = vdwparam[tj+1];
 174                 cexp2            = vdwparam[tj+2];
 175                 rinvsq           = rinv11*rinv11;
 176                 r                = rsq11*rinv11;
 177                 rt               = r*tabscale;
 178                 n0               = rt;
 179                 eps              = rt-n0;
 180                 eps2             = eps*eps;
 181                 nnn              = 4*n0;
 182                 Y                = VFtab[nnn];
 183                 F                = VFtab[nnn+1];
 184                 Geps             = eps*VFtab[nnn+2];
 185                 Heps2            = eps2*VFtab[nnn+3];
 186                 Fp               = F+Geps+Heps2;
 187                 VV               = Y+eps*Fp;
 188                 FF               = Fp+Geps+2.0*Heps2;
 189                 vcoul            = qq*VV;
 190                 fijC             = qq*FF;
 191                 vctot            = vctot + vcoul;
 192                 rinvsix          = rinvsq*rinvsq*rinvsq;
 193                 Vvdw6            = c6*rinvsix;
 194                 br               = cexp2*rsq11*rinv11;
 195                 Vvdwexp          = cexp1*exp(-br);
 196                 Vvdwtot          = Vvdwtot+Vvdwexp-Vvdw6;
 197                 fscal            = (br*Vvdwexp-6.0*Vvdw6)*rinvsq-((fijC)*tabscale)*rinv11;
 198                 fscal *= hybscal;
 199                 tx               = fscal*dx11;
 200                 ty               = fscal*dy11;
 201                 tz               = fscal*dz11;
 202                 fix1             = fix1 + tx;
 203                 fiy1             = fiy1 + ty;
 204                 fiz1             = fiz1 + tz;
 205                 faction[j3+0]    = faction[j3+0] - tx;
 206                 faction[j3+1]    = faction[j3+1] - ty;
 207                 faction[j3+2]    = faction[j3+2] - tz;
 208             }
 209
 210             faction[ii3+0]   = faction[ii3+0] + fix1;
 211             faction[ii3+1]   = faction[ii3+1] + fiy1;
 212             faction[ii3+2]   = faction[ii3+2] + fiz1;
 213             fshift[is3]      = fshift[is3]+fix1;
 214             fshift[is3+1]    = fshift[is3+1]+fiy1;
 215             fshift[is3+2]    = fshift[is3+2]+fiz1;
 216             ggid             = gid[n];
 217             Vc[ggid]         = Vc[ggid] + vctot;
 218             Vvdw[ggid]       = Vvdw[ggid] + Vvdwtot;
 219             ninner           = ninner + nj1 - nj0;
 220         }
 221
 222         nouter           = nouter + nn1 - nn0;
 223     }
 224     while (nn1<nri);
 225
 226     *outeriter       = nouter;
 227     *inneriter       = ninner;
 228 }
 229
 230
 231
 232
 233
 234 /*
 235  * Gromacs nonbonded kernel nb_kernel320_adress_ex
 236  * Coulomb interaction:     Tabulated
 237  * VdW interaction:         Buckingham
 238  * water optimization:      No
 239  * Calculate forces:        yes
 240  */
 241 void nb_kernel320_adress_ex(
 242                     int *           p_nri,
 243                     int *           iinr,
 244                     int *           jindex,
 245                     int *           jjnr,
 246                     int *           shift,
 247                     real *         shiftvec,
 248                     real *         fshift,
 249                     int *           gid,
 250                     real *         pos,
 251                     real *         faction,
 252                     real *         charge,
 253                     real *         p_facel,
 254                     real *         p_krf,
 255                     real *         p_crf,
 256                     real *         Vc,
 257                     int *           type,
 258                     int *           p_ntype,
 259                     real *         vdwparam,
 260                     real *         Vvdw,
 261                     real *         p_tabscale,
 262                     real *         VFtab,
 263                     real *         invsqrta,
 264                     real *         dvda,
 265                     real *         p_gbtabscale,
 266                     real *         GBtab,
 267                     int *           p_nthreads,
 268                     int *           count,
 269                     void *          mtx,
 270                     int *           outeriter,
 271                     int *           inneriter,
 272                     real           force_cap,
 273                     real *         wf)
 274 {
 275     int           nri,ntype,nthreads;
 276     real         facel,krf,crf,tabscale,gbtabscale;
 277     int           n,ii,is3,ii3,k,nj0,nj1,jnr,j3,ggid;
 278     int           nn0,nn1,nouter,ninner;
 279     real         shX,shY,shZ;
 280     real         fscal,tx,ty,tz;
 281     real         rinvsq;
 282     real         iq;
 283     real         qq,vcoul,vctot;
 284     int           nti;
 285     int           tj;
 286     real         rinvsix;
 287     real         Vvdw6,Vvdwtot;
 288     real         r,rt,eps,eps2;
 289     int           n0,nnn;
 290     real         Y,F,Geps,Heps2,Fp,VV;
 291     real         FF;
 292     real         fijC;
 293     real         Vvdwexp,br;
 294     real         ix1,iy1,iz1,fix1,fiy1,fiz1;
 295     real         jx1,jy1,jz1;
 296     real         dx11,dy11,dz11,rsq11,rinv11;
 297     real         c6,cexp1,cexp2;
 298     real         weight_cg1, weight_cg2, weight_product;
 299     real         hybscal;
 300
 301     nri              = *p_nri;
 302     ntype            = *p_ntype;
 303     nthreads         = *p_nthreads;
 304     facel            = *p_facel;
 305     krf              = *p_krf;
 306     crf              = *p_crf;
 307     tabscale         = *p_tabscale;
 308     nouter           = 0;
 309     ninner           = 0;
 310
 311     do
 312     {
 313         #ifdef GMX_THREAD_SHM_FDECOMP
 314         tMPI_Thread_mutex_lock((tMPI_Thread_mutex_t *)mtx);
 315         nn0              = *count;
 316         nn1              = nn0+(nri-nn0)/(2*nthreads)+10;
 317         *count           = nn1;
 318         tMPI_Thread_mutex_unlock((tMPI_Thread_mutex_t *)mtx);
 319         if(nn1>nri) nn1=nri;
 320         #else
 321         nn0 = 0;
 322         nn1 = nri;
 323         #endif
 324
 325         for(n=nn0; (n<nn1); n++)
 326         {
 327             is3              = 3*shift[n];
 328             shX              = shiftvec[is3];
 329             shY              = shiftvec[is3+1];
 330             shZ              = shiftvec[is3+2];
 331             nj0              = jindex[n];
 332             nj1              = jindex[n+1];
 333             ii               = iinr[n];
 334             ii3              = 3*ii;
 335             ix1              = shX + pos[ii3+0];
 336             iy1              = shY + pos[ii3+1];
 337             iz1              = shZ + pos[ii3+2];
 338             iq               = facel*charge[ii];
 339             nti              = 3*ntype*type[ii];
 340             weight_cg1       = wf[ii];
 341             vctot            = 0;
 342             Vvdwtot          = 0;
 343             fix1             = 0;
 344             fiy1             = 0;
 345             fiz1             = 0;
 346
 347             for(k=nj0; (k<nj1); k++)
 348             {
 349                 jnr              = jjnr[k];
 350                 weight_cg2       = wf[jnr];
 351                 weight_product   = weight_cg1*weight_cg2;
 352                 if (weight_product < ALMOST_ZERO) {
 353                 /* force is zero, skip this molecule */
 354                  continue;
 355                 }
 356                 else if (weight_product >= ALMOST_ONE)
 357                 {
 358                        hybscal = 1.0;
 359                 }
 360                 else
 361                 {
 362                    hybscal = weight_product;
 363                 }
 364                 j3               = 3*jnr;
 365                 jx1              = pos[j3+0];
 366                 jy1              = pos[j3+1];
 367                 jz1              = pos[j3+2];
 368                 dx11             = ix1 - jx1;
 369                 dy11             = iy1 - jy1;
 370                 dz11             = iz1 - jz1;
 371                 rsq11            = dx11*dx11+dy11*dy11+dz11*dz11;
 372                 rinv11           = 1.0/sqrt(rsq11);
 373                 qq               = iq*charge[jnr];
 374                 tj               = nti+3*type[jnr];
 375                 c6               = vdwparam[tj];
 376                 cexp1            = vdwparam[tj+1];
 377                 cexp2            = vdwparam[tj+2];
 378                 rinvsq           = rinv11*rinv11;
 379                 r                = rsq11*rinv11;
 380                 rt               = r*tabscale;
 381                 n0               = rt;
 382                 eps              = rt-n0;
 383                 eps2             = eps*eps;
 384                 nnn              = 4*n0;
 385                 Y                = VFtab[nnn];
 386                 F                = VFtab[nnn+1];
 387                 Geps             = eps*VFtab[nnn+2];
 388                 Heps2            = eps2*VFtab[nnn+3];
 389                 Fp               = F+Geps+Heps2;
 390                 VV               = Y+eps*Fp;
 391                 FF               = Fp+Geps+2.0*Heps2;
 392                 vcoul            = qq*VV;
 393                 fijC             = qq*FF;
 394                 vctot            = vctot + vcoul;
 395                 rinvsix          = rinvsq*rinvsq*rinvsq;
 396                 Vvdw6            = c6*rinvsix;
 397                 br               = cexp2*rsq11*rinv11;
 398                 Vvdwexp          = cexp1*exp(-br);
 399                 Vvdwtot          = Vvdwtot+Vvdwexp-Vvdw6;
 400                 fscal            = (br*Vvdwexp-6.0*Vvdw6)*rinvsq-((fijC)*tabscale)*rinv11;
 401                 fscal *= hybscal;
 402                 if(force_cap>0 && (fabs(fscal)> force_cap)){
 403                 fscal=force_cap*fscal/fabs(fscal);
 404                 }
 405                 tx               = fscal*dx11;
 406                 ty               = fscal*dy11;
 407                 tz               = fscal*dz11;
 408                 fix1             = fix1 + tx;
 409                 fiy1             = fiy1 + ty;
 410                 fiz1             = fiz1 + tz;
 411                 faction[j3+0]    = faction[j3+0] - tx;
 412                 faction[j3+1]    = faction[j3+1] - ty;
 413                 faction[j3+2]    = faction[j3+2] - tz;
 414             }
 415
 416             faction[ii3+0]   = faction[ii3+0] + fix1;
 417             faction[ii3+1]   = faction[ii3+1] + fiy1;
 418             faction[ii3+2]   = faction[ii3+2] + fiz1;
 419             fshift[is3]      = fshift[is3]+fix1;
 420             fshift[is3+1]    = fshift[is3+1]+fiy1;
 421             fshift[is3+2]    = fshift[is3+2]+fiz1;
 422             ggid             = gid[n];
 423             Vc[ggid]         = Vc[ggid] + vctot;
 424             Vvdw[ggid]       = Vvdw[ggid] + Vvdwtot;
 425             ninner           = ninner + nj1 - nj0;
 426         }
 427
 428         nouter           = nouter + nn1 - nn0;
 429     }
 430     while (nn1<nri);
 431
 432     *outeriter       = nouter;
 433     *inneriter       = ninner;
 434 }
 435
 436