added partition support in create_lattice function
[physik/posic.git] / moldyn.c
index 4812902..8451c44 100644 (file)
--- a/moldyn.c
+++ b/moldyn.c
 #include <time.h>
 #include <math.h>
 
+#include <fpu_control.h>
+
+#ifdef PARALLEL
+#include <omp.h>
+#endif
+
+#if defined PTHREADS || defined VISUAL_THREAD
+#include <pthread.h>
+#endif
+
 #include "moldyn.h"
 #include "report/report.h"
 
 #include "potentials/tersoff.h"
 #endif
 
+/* pse */
+#define PSE_NAME
+#define PSE_COL
+#include "pse.h"
+#undef PSE_NAME
+#undef PSE_COL
+
+#ifdef PTHREADS
+/* global mutexes */
+pthread_mutex_t *amutex;
+pthread_mutex_t emutex;
+#endif
+
 /*
  * the moldyn functions
  */
@@ -38,6 +61,9 @@ int moldyn_init(t_moldyn *moldyn,int argc,char **argv) {
 
        printf("[moldyn] init\n");
 
+       /* only needed if compiled without -msse2 (float-store prob!) */
+       //fpu_set_rtd();
+
        memset(moldyn,0,sizeof(t_moldyn));
 
        moldyn->argc=argc;
@@ -46,13 +72,28 @@ int moldyn_init(t_moldyn *moldyn,int argc,char **argv) {
        rand_init(&(moldyn->random),NULL,1);
        moldyn->random.status|=RAND_STAT_VERBOSE;
 
+#ifdef PTHREADS
+       pthread_mutex_init(&emutex,NULL);
+#endif
+
        return 0;
 }
 
 int moldyn_shutdown(t_moldyn *moldyn) {
 
+#ifdef PTHREADS
+       int i;
+#endif
+
        printf("[moldyn] shutdown\n");
 
+#ifdef PTHREADS
+       for(i=0;i<moldyn->count;i++)
+               pthread_mutex_destroy(&(amutex[i]));
+       free(amutex);
+       pthread_mutex_destroy(&emutex);
+#endif
+
        moldyn_log_shutdown(moldyn);
        link_cell_shutdown(moldyn);
        rand_close(&(moldyn->random));
@@ -472,7 +513,8 @@ int moldyn_log_shutdown(t_moldyn *moldyn) {
  */
 
 int create_lattice(t_moldyn *moldyn,u8 type,double lc,int element,double mass,
-                   u8 attr,u8 brand,int a,int b,int c,t_3dvec *origin) {
+                   u8 attr,u8 brand,int a,int b,int c,t_3dvec *origin,
+                   u8 p_type,t_part_vals *p_vals) {
 
        int new,count;
        int ret;
@@ -480,6 +522,9 @@ int create_lattice(t_moldyn *moldyn,u8 type,double lc,int element,double mass,
        void *ptr;
        t_atom *atom;
        char name[16];
+#ifdef PTHREADS
+       pthread_mutex_t *mutex;
+#endif
 
        new=a*b*c;
        count=moldyn->count;
@@ -502,6 +547,16 @@ int create_lattice(t_moldyn *moldyn,u8 type,double lc,int element,double mass,
        moldyn->atom=ptr;
        atom=&(moldyn->atom[count]);
 
+#ifdef PTHREADS
+       ptr=realloc(amutex,(count+new)*sizeof(pthread_mutex_t));
+       if(!ptr) {
+               perror("[moldyn] mutex realloc (add atom)");
+               return -1;
+       }
+       amutex=ptr;
+       mutex=&(amutex[count]);
+#endif
+
        /* no atoms on the boundaries (only reason: it looks better!) */
        if(!origin) {
                orig.x=0.5*lc;
@@ -517,21 +572,21 @@ int create_lattice(t_moldyn *moldyn,u8 type,double lc,int element,double mass,
        switch(type) {
                case CUBIC:
                        set_nn_dist(moldyn,lc);
-                       ret=cubic_init(a,b,c,lc,atom,&orig);
+                       ret=cubic_init(a,b,c,lc,atom,&orig,p_type,p_vals);
                        strcpy(name,"cubic");
                        break;
                case FCC:
                        if(!origin)
                                v3_scale(&orig,&orig,0.5);
                        set_nn_dist(moldyn,0.5*sqrt(2.0)*lc);
-                       ret=fcc_init(a,b,c,lc,atom,&orig);
+                       ret=fcc_init(a,b,c,lc,atom,&orig,p_type,p_vals);
                        strcpy(name,"fcc");
                        break;
                case DIAMOND:
                        if(!origin)
                                v3_scale(&orig,&orig,0.25);
                        set_nn_dist(moldyn,0.25*sqrt(3.0)*lc);
-                       ret=diamond_init(a,b,c,lc,atom,&orig);
+                       ret=diamond_init(a,b,c,lc,atom,&orig,p_type,p_vals);
                        strcpy(name,"diamond");
                        break;
                default:
@@ -559,6 +614,9 @@ int create_lattice(t_moldyn *moldyn,u8 type,double lc,int element,double mass,
                atom[ret].tag=count+ret;
                check_per_bound(moldyn,&(atom[ret].r));
                atom[ret].r_0=atom[ret].r;
+#ifdef PTHREADS
+               pthread_mutex_init(&(mutex[ret]),NULL);
+#endif
        }
 
        /* update total system mass */
@@ -584,6 +642,25 @@ int add_atom(t_moldyn *moldyn,int element,double mass,u8 brand,u8 attr,
        }
        moldyn->atom=ptr;
 
+#ifdef LOWMEM_LISTS
+       ptr=realloc(moldyn->lc.subcell->list,(count+1)*sizeof(int));
+       if(!ptr) {
+               perror("[moldyn] list realloc (add atom)");
+               return -1;
+       }
+       moldyn->lc.subcell->list=ptr;
+#endif
+
+#ifdef PTHREADS
+       ptr=realloc(amutex,(count+1)*sizeof(pthread_mutex_t));
+       if(!ptr) {
+               perror("[moldyn] mutex realloc (add atom)");
+               return -1;
+       }
+       amutex=ptr;
+       pthread_mutex_init(&(amutex[count]),NULL);
+#endif
+
        atom=moldyn->atom;
 
        /* initialize new atom */
@@ -634,12 +711,14 @@ int del_atom(t_moldyn *moldyn,int tag) {
 }
 
 /* cubic init */
-int cubic_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
+int cubic_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin,
+               u8 p_type,t_part_vals *p_vals) {
 
        int count;
        t_3dvec r;
        int i,j,k;
        t_3dvec o;
+       t_3dvec dist;
 
        count=0;
        if(origin)
@@ -647,14 +726,39 @@ int cubic_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
        else
                v3_zero(&o);
 
+       /* shift partition values */
+       if(p_type) {
+               p_vals->p.x+=(a*lc)/2.0;
+               p_vals->p.y+=(b*lc)/2.0;
+               p_vals->p.z+=(c*lc)/2.0;
+       }
+
        r.x=o.x;
        for(i=0;i<a;i++) {
                r.y=o.y;
                for(j=0;j<b;j++) {
                        r.z=o.z;
                        for(k=0;k<c;k++) {
+                               switch(p_type) {
+                                       case PART_INSIDE_R:
+                                               v3_sub(&dist,&r,&(p_vals->p));
+                       if(v3_absolute_square(&dist)<(p_vals->r*p_vals->r)) {
                                v3_copy(&(atom[count].r),&r);
                                count+=1;
+                       }
+                                               break;
+                                       case PART_OUTSIDE_R:
+                                               v3_sub(&dist,&r,&(p_vals->p));
+                       if(v3_absolute_square(&dist)>=(p_vals->r*p_vals->r)) {
+                               v3_copy(&(atom[count].r),&r);
+                               count+=1;
+                       }
+                                               break;
+                                       default:        
+                                               v3_copy(&(atom[count].r),&r);
+                                               count+=1;
+                                               break;
+                               }
                                r.z+=lc;
                        }
                        r.y+=lc;
@@ -672,12 +776,14 @@ int cubic_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
 }
 
 /* fcc lattice init */
-int fcc_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
+int fcc_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin,
+             u8 p_type,t_part_vals *p_vals) {
 
        int count;
        int i,j,k,l;
        t_3dvec o,r,n;
        t_3dvec basis[3];
+       t_3dvec dist;
 
        count=0;
        if(origin)
@@ -694,6 +800,13 @@ int fcc_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
        basis[2].y=0.5*lc;
        basis[2].z=0.5*lc;
 
+       /* shift partition values */
+       if(p_type) {
+               p_vals->p.x+=(a*lc)/2.0;
+               p_vals->p.y+=(b*lc)/2.0;
+               p_vals->p.z+=(c*lc)/2.0;
+       }
+
        /* fill up the room */
        r.x=o.x;
        for(i=0;i<a;i++) {
@@ -702,15 +815,51 @@ int fcc_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
                        r.z=o.z;
                        for(k=0;k<c;k++) {
                                /* first atom */
+                               switch(p_type) {
+                                       case PART_INSIDE_R:
+                                               v3_sub(&dist,&r,&(p_vals->p));
+                       if(v3_absolute_square(&dist)<(p_vals->r*p_vals->r)) {
                                v3_copy(&(atom[count].r),&r);
                                count+=1;
-                               r.z+=lc;
+                       }
+                                               break;
+                                       case PART_OUTSIDE_R:
+                                               v3_sub(&dist,&r,&(p_vals->p));
+                       if(v3_absolute_square(&dist)>=(p_vals->r*p_vals->r)) {
+                               v3_copy(&(atom[count].r),&r);
+                               count+=1;
+                       }
+                                               break;
+                                       default:
+                                               v3_copy(&(atom[count].r),&r);
+                                               count+=1;
+                                               break;
+                               }
                                /* the three face centered atoms */
                                for(l=0;l<3;l++) {
                                        v3_add(&n,&r,&basis[l]);
+                                       switch(p_type) {
+                                               case PART_INSIDE_R:
+                       v3_sub(&dist,&n,&(p_vals->p));
+                       if(v3_absolute_square(&dist)<(p_vals->r*p_vals->r)) {
+                               v3_copy(&(atom[count].r),&r);
+                               count+=1;
+                       }
+                                                       break;
+                                               case PART_OUTSIDE_R:
+                       v3_sub(&dist,&n,&(p_vals->p));
+                       if(v3_absolute_square(&dist)>=(p_vals->r*p_vals->r)) {
+                               v3_copy(&(atom[count].r),&r);
+                               count+=1;
+                       }
+                                                       break;
+                                               default:
                                        v3_copy(&(atom[count].r),&n);
                                        count+=1;
+                                                       break;
+                                       }
                                }
+                               r.z+=lc;
                        }
                        r.y+=lc;
                }
@@ -727,12 +876,13 @@ int fcc_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
        return count;
 }
 
-int diamond_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
+int diamond_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin,
+                 u8 p_type,t_part_vals *p_vals) {
 
        int count;
        t_3dvec o;
 
-       count=fcc_init(a,b,c,lc,atom,origin);
+       count=fcc_init(a,b,c,lc,atom,origin,p_type,p_vals);
 
        o.x=0.25*lc;
        o.y=0.25*lc;
@@ -740,7 +890,7 @@ int diamond_init(int a,int b,int c,double lc,t_atom *atom,t_3dvec *origin) {
 
        if(origin) v3_add(&o,&o,origin);
 
-       count+=fcc_init(a,b,c,lc,&atom[count],&o);
+       count+=fcc_init(a,b,c,lc,&atom[count],&o,p_type,p_vals);
 
        return count;
 }
@@ -872,7 +1022,7 @@ int scale_velocity(t_moldyn *moldyn,u8 equi_init) {
                scale*=2.0;
        else
                if(moldyn->pt_scale&T_SCALE_BERENDSEN)
-                       scale=1.0+(scale-1.0)*moldyn->timestep/moldyn->t_tc;
+                       scale=1.0+(scale-1.0)*moldyn->tau/moldyn->t_tc;
        scale=sqrt(scale);
 
        /* velocity scaling */
@@ -1182,8 +1332,8 @@ int scale_volume(t_moldyn *moldyn) {
 
        /* scaling factor */
        if(moldyn->pt_scale&P_SCALE_BERENDSEN) {
-               scale=(moldyn->p_ref-moldyn->p)*moldyn->p_tc*moldyn->timestep;
-               scale=pow(1.0-scale,ONE_THIRD);
+               scale=1.0-(moldyn->p_ref-moldyn->p)*moldyn->p_tc*moldyn->tau;
+               scale=pow(scale,ONE_THIRD);
        }
        else {
                scale=pow(moldyn->p/moldyn->p_ref,ONE_THIRD);
@@ -1277,7 +1427,9 @@ double estimate_time_step(t_moldyn *moldyn,double nn_dist) {
 int link_cell_init(t_moldyn *moldyn,u8 vol) {
 
        t_linkcell *lc;
+#ifndef LOWMEM_LISTS
        int i;
+#endif
 
        lc=&(moldyn->lc);
 
@@ -1292,6 +1444,8 @@ int link_cell_init(t_moldyn *moldyn,u8 vol) {
 
 #ifdef STATIC_LISTS
        lc->subcell=malloc(lc->cells*sizeof(int*));
+#elif LOWMEM_LISTS
+       lc->subcell=malloc(sizeof(t_lowmem_list));
 #else
        lc->subcell=malloc(lc->cells*sizeof(t_list));
 #endif
@@ -1302,12 +1456,16 @@ int link_cell_init(t_moldyn *moldyn,u8 vol) {
        }
 
        if(lc->cells<27)
-               printf("[moldyn] FATAL: less then 27 subcells!\n");
+               printf("[moldyn] FATAL: less then 27 subcells! (%d)\n",
+                       lc->cells);
 
        if(vol) {
 #ifdef STATIC_LISTS
                printf("[moldyn] initializing 'static' linked cells (%d)\n",
                       lc->cells);
+#elif LOWMEM_LISTS
+               printf("[moldyn] initializing 'lowmem' linked cells (%d)\n",
+                      lc->cells);
 #else
                printf("[moldyn] initializing 'dynamic' linked cells (%d)\n",
                       lc->cells);
@@ -1331,6 +1489,17 @@ int link_cell_init(t_moldyn *moldyn,u8 vol) {
                               i,lc->subcell[0],lc->subcell);
                */
        }
+#elif LOWMEM_LISTS
+       lc->subcell->head=malloc(lc->cells*sizeof(int));
+       if(lc->subcell->head==NULL) {
+               perror("[moldyn] head init (malloc)");
+               return -1;
+       }
+       lc->subcell->list=malloc(moldyn->count*sizeof(int));
+       if(lc->subcell->list==NULL) {
+               perror("[moldyn] list init (malloc)");
+               return -1;
+       }
 #else
        for(i=0;i<lc->cells;i++)
                list_init_f(&(lc->subcell[i]));
@@ -1345,22 +1514,26 @@ int link_cell_init(t_moldyn *moldyn,u8 vol) {
 int link_cell_update(t_moldyn *moldyn) {
 
        int count,i,j,k;
-       int nx,ny;
+       int nx,nxy;
        t_atom *atom;
        t_linkcell *lc;
 #ifdef STATIC_LISTS
        int p;
+#elif LOWMEM_LISTS
+       int p;
 #endif
 
        atom=moldyn->atom;
        lc=&(moldyn->lc);
 
        nx=lc->nx;
-       ny=lc->ny;
+       nxy=nx*lc->ny;
 
        for(i=0;i<lc->cells;i++)
 #ifdef STATIC_LISTS
-               memset(lc->subcell[i],0,(MAX_ATOMS_PER_LIST+1)*sizeof(int));
+               memset(lc->subcell[i],-1,(MAX_ATOMS_PER_LIST+1)*sizeof(int));
+#elif LOWMEM_LISTS
+               lc->subcell->head[i]=-1;
 #else
                list_destroy_f(&(lc->subcell[i]));
 #endif
@@ -1372,7 +1545,7 @@ int link_cell_update(t_moldyn *moldyn) {
        
 #ifdef STATIC_LISTS
                p=0;
-               while(lc->subcell[i+j*nx+k*nx*ny][p]!=0)
+               while(lc->subcell[i+j*nx+k*nxy][p]!=-1)
                        p++;
 
                if(p>=MAX_ATOMS_PER_LIST) {
@@ -1380,9 +1553,13 @@ int link_cell_update(t_moldyn *moldyn) {
                        return -1;
                }
 
-               lc->subcell[i+j*nx+k*nx*ny][p]=count;
+               lc->subcell[i+j*nx+k*nxy][p]=count;
+#elif LOWMEM_LISTS
+               p=i+j*nx+k*nxy;
+               lc->subcell->list[count]=lc->subcell->head[p];
+               lc->subcell->head[p]=count;
 #else
-               list_add_immediate_f(&(lc->subcell[i+j*nx+k*nx*ny]),
+               list_add_immediate_f(&(lc->subcell[i+j*nx+k*nxy]),
                                     &(atom[count]));
                /*
                if(j==0&&k==0)
@@ -1398,6 +1575,8 @@ int link_cell_update(t_moldyn *moldyn) {
 int link_cell_neighbour_index(t_moldyn *moldyn,int i,int j,int k,
 #ifdef STATIC_LISTS
                               int **cell
+#elif LOWMEM_LISTS
+                              int *cell
 #else
                               t_list *cell
 #endif
@@ -1423,7 +1602,11 @@ int link_cell_neighbour_index(t_moldyn *moldyn,int i,int j,int k,
                printf("[moldyn] WARNING: lcni %d/%d %d/%d %d/%d\n",
                       i,nx,j,ny,k,nz);
 
+#ifndef LOWMEM_LISTS
        cell[0]=lc->subcell[i+j*nx+k*a];
+#else
+       cell[0]=lc->subcell->head[i+j*nx+k*a];
+#endif
        for(ci=-1;ci<=1;ci++) {
                bx=0;
                x=i+ci;
@@ -1447,10 +1630,19 @@ int link_cell_neighbour_index(t_moldyn *moldyn,int i,int j,int k,
                                }
                                if(!(ci|cj|ck)) continue;
                                if(bx|by|bz) {
+#ifndef LOWMEM_LISTS
                                        cell[--count2]=lc->subcell[x+y*nx+z*a];
+#else
+                               cell[--count2]=lc->subcell->head[x+y*nx+z*a];
+#endif
+                                       
                                }
                                else {
+#ifndef LOWMEM_LISTS
                                        cell[count1++]=lc->subcell[x+y*nx+z*a];
+#else
+                               cell[count1++]=lc->subcell->head[x+y*nx+z*a];
+#endif
                                }
                        }
                }
@@ -1463,11 +1655,19 @@ int link_cell_neighbour_index(t_moldyn *moldyn,int i,int j,int k,
 
 int link_cell_shutdown(t_moldyn *moldyn) {
 
+#ifndef LOWMEM_LISTS
        int i;
+#endif
        t_linkcell *lc;
 
        lc=&(moldyn->lc);
 
+#if LOWMEM_LISTS
+       free(lc->subcell->head);
+       free(lc->subcell->list);
+
+#else
+
        for(i=0;i<lc->cells;i++) {
 #ifdef STATIC_LISTS
                free(lc->subcell[i]);
@@ -1476,6 +1676,7 @@ int link_cell_shutdown(t_moldyn *moldyn) {
                list_destroy_f(&(lc->subcell[i]));
 #endif
        }
+#endif
 
        free(lc->subcell);
 
@@ -1544,6 +1745,17 @@ int moldyn_integrate(t_moldyn *moldyn) {
        struct timeval t1,t2;
        //double tp;
 
+#ifdef VISUAL_THREAD
+       u8 first,change;
+       pthread_t io_thread;
+       int ret;
+       t_moldyn md_copy;
+       t_atom *atom_copy;
+
+       first=1;
+       change=0;
+#endif
+
        sched=&(moldyn->schedule);
        atom=moldyn->atom;
 
@@ -1592,6 +1804,21 @@ int moldyn_integrate(t_moldyn *moldyn) {
        /* debugging, ignore */
        moldyn->debug=0;
 
+       /* zero & init moldyn copy */
+#ifdef VISUAL_THREAD
+       memset(&md_copy,0,sizeof(t_moldyn));
+       atom_copy=malloc(moldyn->count*sizeof(t_atom));
+       if(atom_copy==NULL) {
+               perror("[moldyn] malloc atom copy (init)");
+               return -1;
+       }
+#endif
+
+#ifdef PTHREADS
+       printf("##################\n");
+       printf("# USING PTHREADS #\n");
+       printf("##################\n");
+#endif
        /* tell the world */
        printf("[moldyn] integration start, go get a coffee ...\n");
 
@@ -1676,7 +1903,7 @@ int moldyn_integrate(t_moldyn *moldyn) {
                }
                if(s) {
                        if(!(moldyn->total_steps%s)) {
-                               snprintf(dir,128,"%s/s-%07.f.save",
+                               snprintf(dir,128,"%s/s-%08.f.save",
                                         moldyn->vlsdir,moldyn->time);
                                fd=open(dir,O_WRONLY|O_TRUNC|O_CREAT,
                                        S_IRUSR|S_IWUSR);
@@ -1691,16 +1918,45 @@ int moldyn_integrate(t_moldyn *moldyn) {
                }
                if(a) {
                        if(!(moldyn->total_steps%a)) {
+#ifdef VISUAL_THREAD
+       /* check whether thread has not terminated yet */
+       if(!first) {
+               ret=pthread_join(io_thread,NULL);
+       }
+       first=0;
+       /* prepare and start thread */
+       if(moldyn->count!=md_copy.count) {
+               free(atom_copy);
+               change=1;
+       }
+       memcpy(&md_copy,moldyn,sizeof(t_moldyn));
+       if(change) {
+               atom_copy=malloc(moldyn->count*sizeof(t_atom));
+               if(atom_copy==NULL) {
+                       perror("[moldyn] malloc atom copy (change)");
+                       return -1;
+               }
+       }
+       md_copy.atom=atom_copy;
+       memcpy(atom_copy,moldyn->atom,moldyn->count*sizeof(t_atom));
+       change=0;
+       ret=pthread_create(&io_thread,NULL,visual_atoms,&md_copy);
+       if(ret) {
+               perror("[moldyn] create visual atoms thread\n");
+               return -1;
+       }
+#else
                                visual_atoms(moldyn);
+#endif
                        }
                }
 
                /* display progress */
-               //if(!(moldyn->total_steps%10)) {
+               if(!(i%10)) {
                        /* get current time */
                        gettimeofday(&t2,NULL);
 
-printf("\rsched:%d, steps:%d/%d, T:%4.1f/%4.1f P:%4.1f/%4.1f V:%6.1f (%d)",
+printf("sched:%d, steps:%d/%d, T:%4.1f/%4.1f P:%4.1f/%4.1f V:%6.1f (%d)\n",
        sched->count,i,moldyn->total_steps,
        moldyn->t,moldyn->t_avg,
        moldyn->p/BAR,moldyn->p_avg/BAR,
@@ -1712,7 +1968,7 @@ printf("\rsched:%d, steps:%d/%d, T:%4.1f/%4.1f P:%4.1f/%4.1f V:%6.1f (%d)",
 
                        /* copy over time */
                        t1=t2;
-               //}
+               }
 
                /* increase absolute time */
                moldyn->time+=moldyn->tau;
@@ -1774,7 +2030,11 @@ int velocity_verlet(t_moldyn *moldyn) {
        link_cell_update(moldyn);
 
        /* forces depending on chosen potential */
+#ifndef ALBE_FAST
        potential_force_calc(moldyn);
+#else
+       albe_potential_force_calc(moldyn);
+#endif
 
        for(i=0;i<count;i++) {
                /* check whether fixed atom */
@@ -1807,6 +2067,9 @@ int potential_force_calc(t_moldyn *moldyn) {
        int *neighbour_i[27];
        int p,q;
        t_atom *atom;
+#elif LOWMEM_LISTS
+       int neighbour_i[27];
+       int p,q;
 #else
        t_list neighbour_i[27];
        t_list neighbour_i2[27];
@@ -1829,6 +2092,10 @@ int potential_force_calc(t_moldyn *moldyn) {
        memset(&(moldyn->gvir),0,sizeof(t_virial));
 
        /* reset force, site energy and virial of every atom */
+#ifdef PARALLEL
+       i=omp_get_thread_num();
+       #pragma omp parallel for private(virial)
+#endif
        for(i=0;i<count;i++) {
 
                /* reset force */
@@ -1879,22 +2146,17 @@ int potential_force_calc(t_moldyn *moldyn) {
 #ifdef STATIC_LISTS
                                p=0;
 
-                               while(neighbour_i[j][p]!=0) {
+                               while(neighbour_i[j][p]!=-1) {
 
                                        jtom=&(atom[neighbour_i[j][p]]);
                                        p++;
+#elif LOWMEM_LISTS
+                               p=neighbour_i[j];
 
-                                       if(jtom==&(itom[i]))
-                                               continue;
+                               while(p!=-1) {
 
-                                       if((jtom->attr&ATOM_ATTR_2BP)&
-                                          (itom[i].attr&ATOM_ATTR_2BP)) {
-                                               moldyn->func2b(moldyn,
-                                                              &(itom[i]),
-                                                              jtom,
-                                                              bc_ij);
-                                       }
-                               }
+                                       jtom=&(itom[p]);
+                                       p=lc->subcell->list[p];
 #else
                                this=&(neighbour_i[j]);
                                list_reset_f(this);
@@ -1904,6 +2166,7 @@ int potential_force_calc(t_moldyn *moldyn) {
 
                                do {
                                        jtom=this->current->data;
+#endif
 
                                        if(jtom==&(itom[i]))
                                                continue;
@@ -1915,6 +2178,11 @@ int potential_force_calc(t_moldyn *moldyn) {
                                                               jtom,
                                                               bc_ij);
                                        }
+#ifdef STATIC_LISTS
+                               }
+#elif LOWMEM_LISTS
+                               }
+#else
                                } while(list_next_f(this)!=L_NO_NEXT_ELEMENT);
 #endif
 
@@ -1929,6 +2197,8 @@ int potential_force_calc(t_moldyn *moldyn) {
                /* copy the neighbour lists */
 #ifdef STATIC_LISTS
                /* no copy needed for static lists */
+#elif LOWMEM_LISTS
+               /* no copy needed for lowmem lists */
 #else
                memcpy(neighbour_i2,neighbour_i,27*sizeof(t_list));
 #endif
@@ -1940,10 +2210,17 @@ int potential_force_calc(t_moldyn *moldyn) {
 #ifdef STATIC_LISTS
                        p=0;
 
-                       while(neighbour_i[j][p]!=0) {
+                       while(neighbour_i[j][p]!=-1) {
 
                                jtom=&(atom[neighbour_i[j][p]]);
                                p++;
+#elif LOWMEM_LISTS
+                               p=neighbour_i[j];
+
+                               while(p!=-1) {
+
+                                       jtom=&(itom[p]);
+                                       p=lc->subcell->list[p];
 #else
                        this=&(neighbour_i[j]);
                        list_reset_f(this);
@@ -1984,10 +2261,17 @@ int potential_force_calc(t_moldyn *moldyn) {
 #ifdef STATIC_LISTS
                                        q=0;
 
-                                       while(neighbour_i[j][q]!=0) {
+                                       while(neighbour_i[k][q]!=-1) {
 
                                                ktom=&(atom[neighbour_i[k][q]]);
                                                q++;
+#elif LOWMEM_LISTS
+                                       q=neighbour_i[k];
+
+                                       while(q!=-1) {
+
+                                               ktom=&(itom[q]);
+                                               q=lc->subcell->list[q];
 #else
                                        that=&(neighbour_i2[k]);
                                        list_reset_f(that);
@@ -2013,8 +2297,11 @@ int potential_force_calc(t_moldyn *moldyn) {
                                                                  jtom,
                                                                  ktom,
                                                                  bc_ik|bc_ij);
+
 #ifdef STATIC_LISTS
                                        }
+#elif LOWMEM_LISTS
+                                       }
 #else
                                        } while(list_next_f(that)!=\
                                                L_NO_NEXT_ELEMENT);
@@ -2039,10 +2326,17 @@ int potential_force_calc(t_moldyn *moldyn) {
 #ifdef STATIC_LISTS
                                        q=0;
 
-                                       while(neighbour_i[j][q]!=0) {
+                                       while(neighbour_i[k][q]!=-1) {
 
                                                ktom=&(atom[neighbour_i[k][q]]);
                                                q++;
+#elif LOWMEM_LISTS
+                                       q=neighbour_i[k];
+
+                                       while(q!=-1) {
+
+                                               ktom=&(itom[q]);
+                                               q=lc->subcell->list[q];
 #else
                                        that=&(neighbour_i2[k]);
                                        list_reset_f(that);
@@ -2071,6 +2365,8 @@ int potential_force_calc(t_moldyn *moldyn) {
 
 #ifdef STATIC_LISTS
                                        }
+#elif LOWMEM_LISTS
+                                       }
 #else
                                        } while(list_next_f(that)!=\
                                                L_NO_NEXT_ELEMENT);
@@ -2088,6 +2384,8 @@ int potential_force_calc(t_moldyn *moldyn) {
                                }
 #ifdef STATIC_LISTS
                        }
+#elif LOWMEM_LISTS
+                       }
 #else
                        } while(list_next_f(this)!=L_NO_NEXT_ELEMENT);
 #endif
@@ -2114,6 +2412,9 @@ int potential_force_calc(t_moldyn *moldyn) {
 #endif
 
        /* some postprocessing */
+#ifdef PARALLEL
+       #pragma omp parallel for
+#endif
        for(i=0;i<count;i++) {
                /* calculate global virial */
                moldyn->gvir.xx+=itom[i].r.x*itom[i].f.x;
@@ -2326,13 +2627,16 @@ int process_2b_bonds(t_moldyn *moldyn,void *data,
 #ifdef STATIC_LISTS
        int *neighbour[27];
        int p;
+#elif LOWMEM_LISTS
+       int neighbour[27];
+       int p;
 #else
        t_list neighbour[27];
+       t_list *this;
 #endif
        u8 bc;
        t_atom *itom,*jtom;
        int i,j;
-       t_list *this;
 
        lc=&(moldyn->lc);
        itom=moldyn->atom;
@@ -2352,10 +2656,17 @@ int process_2b_bonds(t_moldyn *moldyn,void *data,
 #ifdef STATIC_LISTS
                        p=0;
 
-                       while(neighbour[j][p]!=0) {
+                       while(neighbour[j][p]!=-1) {
 
                                jtom=&(moldyn->atom[neighbour[j][p]]);
                                p++;
+#elif LOWMEM_LISTS
+                       p=neighbour[j];
+
+                       while(p!=-1) {
+
+                               jtom=&(itom[p]);
+                               p=lc->subcell->list[p];
 #else
                        this=&(neighbour[j]);
                        list_reset_f(this);
@@ -2373,6 +2684,8 @@ int process_2b_bonds(t_moldyn *moldyn,void *data,
 
 #ifdef STATIC_LISTS
                        }
+#elif LOWMEM_LISTS
+                       }
 #else
                        } while(list_next_f(this)!=L_NO_NEXT_ELEMENT);
 #endif
@@ -2383,6 +2696,84 @@ int process_2b_bonds(t_moldyn *moldyn,void *data,
 
 }
 
+/*
+ * function to find neighboured atoms
+ */
+
+int process_neighbours(t_moldyn *moldyn,void *data,t_atom *atom,
+                       int (*process)(t_moldyn *moldyn,t_atom *atom,t_atom *natom,
+                                     void *data,u8 bc)) {
+
+       t_linkcell *lc;
+#ifdef STATIC_LISTS
+       int *neighbour[27];
+       int p;
+#elif LOWMEM_LISTS
+       int neighbour[27];
+       int p;
+#else
+       t_list neighbour[27];
+       t_list *this;
+#endif
+       u8 bc;
+       t_atom *natom;
+       int j;
+
+       lc=&(moldyn->lc);
+       
+       /* neighbour indexing */
+       link_cell_neighbour_index(moldyn,
+                                 (atom->r.x+moldyn->dim.x/2)/lc->x,
+                                 (atom->r.y+moldyn->dim.y/2)/lc->x,
+                                 (atom->r.z+moldyn->dim.z/2)/lc->x,
+                                 neighbour);
+
+       for(j=0;j<27;j++) {
+
+               bc=(j<lc->dnlc)?0:1;
+
+#ifdef STATIC_LISTS
+               p=0;
+
+               while(neighbour[j][p]!=-1) {
+
+                       natom=&(moldyn->atom[neighbour[j][p]]);
+                       p++;
+#elif LOWMEM_LISTS
+               p=neighbour[j];
+
+               while(p!=-1) {
+
+                       natom=&(moldyn->atom[p]);
+                       p=lc->subcell->list[p];
+#else
+               this=&(neighbour[j]);
+               list_reset_f(this);
+
+               if(this->start==NULL)
+                       continue;
+
+               do {
+
+                       natom=this->current->data;
+#endif
+
+                       /* process bond */
+                       process(moldyn,atom,natom,data,bc);
+
+#ifdef STATIC_LISTS
+               }
+#elif LOWMEM_LISTS
+               }
+#else
+               } while(list_next_f(this)!=L_NO_NEXT_ELEMENT);
+#endif
+       }
+
+       return 0;
+
+}
+
 /*
  * post processing functions
  */
@@ -2709,7 +3100,11 @@ int visual_bonds_process(t_moldyn *moldyn,t_atom *itom,t_atom *jtom,
        return 0;
 }
 
+#ifdef VISUAL_THREAD
+void *visual_atoms(void *ptr) {
+#else
 int visual_atoms(t_moldyn *moldyn) {
+#endif
 
        int i;
        char file[128+64];
@@ -2718,6 +3113,11 @@ int visual_atoms(t_moldyn *moldyn) {
        t_visual *v;
        t_atom *atom;
        t_vb vb;
+#ifdef VISUAL_THREAD
+       t_moldyn *moldyn;
+
+       moldyn=ptr;
+#endif
 
        v=&(moldyn->vis);
        dim.x=v->dim.x;
@@ -2727,17 +3127,19 @@ int visual_atoms(t_moldyn *moldyn) {
 
        help=(dim.x+dim.y);
 
-       sprintf(file,"%s/atomic_conf_%07.f.xyz",v->fb,moldyn->time);
+       sprintf(file,"%s/atomic_conf_%08.f.xyz",v->fb,moldyn->time);
        vb.fd=open(file,O_WRONLY|O_CREAT|O_TRUNC,S_IRUSR|S_IWUSR);
        if(vb.fd<0) {
                perror("open visual save file fd");
+#ifndef VISUAL_THREAD
                return -1;
+#endif
        }
 
        /* write the actual data file */
 
        // povray header
-       dprintf(vb.fd,"# [P] %d %07.f <%f,%f,%f>\n",
+       dprintf(vb.fd,"# [P] %d %08.f <%f,%f,%f>\n",
                moldyn->count,moldyn->time,help/40.0,help/40.0,-0.8*help);
 
        // atomic configuration
@@ -2751,7 +3153,9 @@ int visual_atoms(t_moldyn *moldyn) {
                                                    atom[i].ekin);
        
        // bonds between atoms
+#ifndef VISUAL_THREAD
        process_2b_bonds(moldyn,&vb,visual_bonds_process);
+#endif
        
        // boundaries
        if(dim.x) {
@@ -2797,6 +3201,32 @@ int visual_atoms(t_moldyn *moldyn) {
 
        close(vb.fd);
 
+#ifdef VISUAL_THREAD
+       pthread_exit(NULL);
+
+}
+#else
+
+       return 0;
+}
+#endif
+
+/*
+ * fpu cntrol functions
+ */
+
+// set rounding to double (eliminates -ffloat-store!)
+int fpu_set_rtd(void) {
+
+       fpu_control_t ctrl;
+
+       _FPU_GETCW(ctrl);
+
+       ctrl&=~_FPU_EXTENDED;
+       ctrl|=_FPU_DOUBLE;
+
+       _FPU_SETCW(ctrl);
+
        return 0;
 }