multicore.cpp

Go to the documentation of this file.

/* included by treeset.cpp */
/*#include <../../nrnconf.h>*/
/*#include <multicore.h>*/
#include <nrnpthread.h>
#include <nrnmpi.h>


/*
Now that threads have taken over the actual_v, v_node, etc, it might
be a good time to regularize the method of freeing, allocating, and
updating those arrays. To recapitulate the history, Node used to
be the structure that held direct values for v, area, d, rhs, etc.
That continued to hold for the cray vectorization project which
introduced v_node, v_parent, memb_list. Cache efficiency introduced
actual_v, actual_area, actual_d, etc and the Node started pointing
into those arrays. Additional nodes after allocation required updating
pointers to v and area since those arrays were freed and reallocated.
Now, the threads hold all these arrays and we want to update them
properly under the circumstances of changing topology, changing
number of threads, and changing distribution of cells on threads.
Note there are no longer global versions of any of these arrays.
We do not want to update merely due to a change in area. Recently
we have dealt with diam, area, ri on a section basis. We generally
desire an update just before a simulation when the efficient
structures are necessary. This is reasonably well handled by the
v_structure_change flag which historically freed and reallocated
v_node and v_parent and, just before this comment,
ended up setting the NrnThread tml. This makes most of the old
memb_list vestigial and we now got rid of it except for
the artificial cells (and it is possibly not really necessary there).
Switching between sparse and tree matrix just cause freeing and
reallocation of actual_rhs.

If we can get the freeing, reallocation, and pointer update correct
for _actual_v, I am guessing everything else can be dragged along with
it. We have two major cases, call to pc.nthread and change in
model structure. We want to use Node* as much as possible and defer
the handling of v_structure_change as long as possible.
*/

#define CACHELINE_ALLOC(name,type,size) name = (type*)nrn_cacheline_alloc((void**)&name, size*sizeof(type))
#define CACHELINE_CALLOC(name,type,size) name = (type*)nrn_cacheline_calloc((void**)&name, size, sizeof(type))

int nrn_nthread;
NrnThread* nrn_threads;
void (*nrn_mk_transfer_thread_data_)();

static int busywait_;
static int busywait_main_;
extern void nrn_thread_error(const char*);
extern void nrn_threads_free();
extern void nrn_old_thread_save();
extern double nrn_timeus();

static int nrn_thread_parallel_;
void nrn_mk_table_check();
static int table_check_cnt_;
static Datum* table_check_;
static int allow_busywait_;

/* linux specfic for performance testing */
/* eventually will be removed */
#define BENCHMARKING 0
#if BENCHMARKING
/* for rdtscll() */
#include <asm/msr.h>
#define BENCHDECLARE unsigned long t1;
#define BENCHBEGIN(arg) if (t_[arg] < t1_[arg] + BSIZE) {rdtscl(t1); *(t_[arg]++) = t1;}
#define BENCHADD(arg) BENCHBEGIN(arg)
#define WAIT wait_for_workers_timeit
#define CPU_MHZ 3192
#define BSIZE 200000
#define BS 10
static unsigned long bcnt_, bcnt1_;
static unsigned long t1_[BS][BSIZE], *t_[BS];
#else
#define BENCHDECLARE /**/
#define BENCHBEGIN(arg) /**/
#define BENCHADD(arg) /**/
#define WAIT wait_for_workers
#define BS 0
#endif

static void* nulljob(NrnThread* nt) {
    return nullptr;
}

int nrn_inthread_;
#if USE_PTHREAD

#include <pthread.h>
#include <sched.h> /* for sched_setaffinity */

/* abort if using threads and a call to malloc is unprotected */
#define use_malloc_hook 0
#if use_malloc_hook
#include <malloc.h>

static int nrn_malloc_protected_;
static void my_init_hook();
static void *(*old_malloc_hook) (size_t, const void*);
static void *(*old_memalign_hook) (size_t, size_t, const void*);
static void *(*old_realloc_hook) (void*, size_t, const void*);
static void (*old_free_hook) (void*, const void*);
static void *my_malloc_hook (size_t, const void*);
static void *my_memalign_hook (size_t, size_t, const void*);
static void *my_realloc_hook (void*, size_t, const void*);
static void my_free_hook (void*, const void*);
void (*__malloc_initialize_hook)(void) = my_init_hook;

static void* my_malloc_hook(size_t size, const void* caller) {
    void* result;
    if (nrn_inthread_ && !nrn_malloc_protected_) {
        abort();
    }
    __malloc_hook = old_malloc_hook;
    __memalign_hook = old_memalign_hook;
    __realloc_hook = old_realloc_hook;
    __free_hook = old_free_hook;
    result = malloc(size);
    old_malloc_hook = __malloc_hook;
    old_memalign_hook = __memalign_hook;
    old_realloc_hook = __realloc_hook;
    old_free_hook = __free_hook;
    __malloc_hook = my_malloc_hook;
    __memalign_hook = my_memalign_hook;
    __realloc_hook = my_realloc_hook;
    __free_hook = my_free_hook;
    return result;
}
static void* my_memalign_hook(size_t alignment, size_t size, const void* caller) {
    void* result;
    if (nrn_inthread_ && !nrn_malloc_protected_) {
        abort();
    }
    __malloc_hook = old_malloc_hook;
    __memalign_hook = old_memalign_hook;
    __realloc_hook = old_realloc_hook;
    __free_hook = old_free_hook;
    result = memalign(alignment, size);
    old_malloc_hook = __malloc_hook;
    old_memalign_hook = __memalign_hook;
    old_realloc_hook = __realloc_hook;
    old_free_hook = __free_hook;
    __malloc_hook = my_malloc_hook;
    __memalign_hook = my_memalign_hook;
    __realloc_hook = my_realloc_hook;
    __free_hook = my_free_hook;
    return result;
}
static void* my_realloc_hook(void* ptr, size_t size, const void* caller) {
    void* result;
    if (nrn_inthread_ && !nrn_malloc_protected_) {
        abort();
    }
    __malloc_hook = old_malloc_hook;
    __memalign_hook = old_memalign_hook;
    __realloc_hook = old_realloc_hook;
    __free_hook = old_free_hook;
    result = realloc(ptr, size);
    old_malloc_hook = __malloc_hook;
    old_memalign_hook = __memalign_hook;
    old_realloc_hook = __realloc_hook;
    old_free_hook = __free_hook;
    __malloc_hook = my_malloc_hook;
    __memalign_hook = my_memalign_hook;
    __realloc_hook = my_realloc_hook;
    __free_hook = my_free_hook;
    return result;
}
static void my_free_hook(void* ptr, const void* caller) {
    if (nrn_inthread_ && !nrn_malloc_protected_) {
        abort();
    }
    __malloc_hook = old_malloc_hook;
    __memalign_hook = old_memalign_hook;
    __realloc_hook = old_realloc_hook;
    __free_hook = old_free_hook;
    free(ptr);
    old_malloc_hook = __malloc_hook;
    old_memalign_hook = __memalign_hook;
    old_realloc_hook = __realloc_hook;
    old_free_hook = __free_hook;
    __malloc_hook = my_malloc_hook;
    __memalign_hook = my_memalign_hook;
    __realloc_hook = my_realloc_hook;
    __free_hook = my_free_hook;
}
static void my_init_hook() {
    static int installed = 0;
    if (installed) { return; }
    installed = 1;
    old_malloc_hook = __malloc_hook;
    __malloc_hook = my_malloc_hook;
    old_memalign_hook = __memalign_hook;
    __memalign_hook = my_memalign_hook;
    old_realloc_hook = __realloc_hook;
    __realloc_hook = my_realloc_hook;
    old_free_hook = __free_hook;
    __free_hook = my_free_hook;
}
#endif

static int interpreter_locked;
static pthread_mutex_t interpreter_lock_;
static pthread_mutex_t* _interpreter_lock;

static pthread_mutex_t nmodlmutex_;
pthread_mutex_t* _nmodlmutex;

static pthread_mutex_t nrn_malloc_mutex_;
static pthread_mutex_t* _nrn_malloc_mutex;

extern "C" void nrn_malloc_lock() {
    if (_nrn_malloc_mutex) {
        pthread_mutex_lock(_nrn_malloc_mutex);
#if use_malloc_hook
        nrn_malloc_protected_ = 1;
#endif
    }
}

extern "C" void nrn_malloc_unlock() {
    if (_nrn_malloc_mutex) {
#if use_malloc_hook
        nrn_malloc_protected_ = 0;
#endif
        pthread_mutex_unlock(_nrn_malloc_mutex);
    }
}

/* when PERMANENT is 0, we avoid false warnings with helgrind, but a bit slower */
/* when 0, create/join instead of wait on condition. */
#ifndef PERMANENT
#define PERMANENT 1
#endif

typedef volatile struct {
        int flag;
    int thread_id;
        /* for nrn_solve etc.*/
        void* (*job)(NrnThread*);
} slave_conf_t;

static pthread_cond_t* cond;
static pthread_mutex_t* mut;
static pthread_t* slave_threads;
static slave_conf_t* wc;

static void wait_for_workers() {
    int i;
    for (i=1; i < nrn_nthread; ++i) {
#if PERMANENT
        if (busywait_main_) {
        while (wc[i].flag != 0){;}
        }else{
        pthread_mutex_lock(mut + i);
        while (wc[i].flag != 0) {
            pthread_cond_wait(cond + i, mut + i);
        }
        pthread_mutex_unlock(mut + i);
        }
#else
        pthread_join(slave_threads[i], nullptr);
#endif
    }
}

static void wait_for_workers_timeit() {
    BENCHDECLARE
    BENCHBEGIN(BS-2)
    wait_for_workers();
    BENCHADD(BS-1)
}

static void send_job_to_slave(int i, void* (*job)(NrnThread*)) {
#if PERMANENT
    pthread_mutex_lock(mut + i);
    wc[i].job = job;
    wc[i].flag = 1;
    pthread_cond_signal(cond + i);
    pthread_mutex_unlock(mut + i);
#else
    pthread_create(slave_threads + i, nullptr, (void*(*)(void*))job, (void*)(nrn_threads + i));
#endif
}

void setaffinity(int i) {
    int mask;
return;
#if 0
    cpu_set_t mask;
    CPU_ZERO(&mask);
    CPU_SET(i, &mask);
    mask = (1 << i);
    sched_setaffinity(0, 4, &mask);
#endif
}

static void* slave_main(void* arg) {
    slave_conf_t* my_wc = (slave_conf_t*)arg;
    pthread_mutex_t *my_mut = mut + my_wc->thread_id;
    pthread_cond_t *my_cond = cond + my_wc->thread_id;
    BENCHDECLARE
#if BENCHMARKING
    unsigned long* t_[BS];
    int a1, a2;
    a1 = my_wc->thread_id;
    a2 = my_wc->thread_id + nrn_nthread;
    t_[a1] = t1_[a1];
    t_[a2] = t1_[a2];
#endif
    setaffinity(my_wc->thread_id);

    for(;;) {
        if (busywait_) {
            while(my_wc->flag == 0) {;}
            if (my_wc->flag == 1) {
                BENCHBEGIN(a1)
                (*my_wc->job)(nrn_threads + my_wc->thread_id);
                BENCHADD(a2)
            }else{
                return nullptr;
            }
            my_wc->flag = 0;
            pthread_cond_signal(my_cond);
        }else{
        pthread_mutex_lock(my_mut);
        while (my_wc->flag == 0) {
            pthread_cond_wait(my_cond, my_mut);
        }
        pthread_mutex_unlock(my_mut);
        pthread_mutex_lock(my_mut);
        if (my_wc->flag == 1) {
            pthread_mutex_unlock(my_mut);
            BENCHBEGIN(a1)
            (*my_wc->job)(nrn_threads + my_wc->thread_id);
            BENCHADD(a2)
        }else{
            pthread_mutex_unlock(my_mut);
            return nullptr;
        }
        pthread_mutex_lock(my_mut);
        my_wc->flag = 0;
        pthread_cond_signal(my_cond);
        pthread_mutex_unlock(my_mut);
        }
    }
    return nullptr;
}

static void threads_create_pthread(){
#if NRNMPI
    if (nrn_nthread > 1 && nrnmpi_numprocs > 1 && nrn_cannot_use_threads_and_mpi == 1) {
    if (nrnmpi_myid == 0) { printf("This MPI is not threadsafe so pthreads are disabled.\n"); }
    nrn_thread_parallel_ = 0;
    return;
    }
#endif
    setaffinity(nrnmpi_myid);
    if (nrn_nthread > 1) {
    int i;
#if PERMANENT
    CACHELINE_ALLOC(wc, slave_conf_t, nrn_nthread);
    slave_threads = (pthread_t *)emalloc(sizeof(pthread_t)*nrn_nthread);
    cond = (pthread_cond_t *)emalloc(sizeof(pthread_cond_t)*nrn_nthread);
    mut = (pthread_mutex_t *)emalloc(sizeof(pthread_mutex_t)*nrn_nthread);
    for (i=1; i < nrn_nthread; ++i) {
        wc[i].flag = 0;
        wc[i].thread_id = i;
        pthread_cond_init(cond + i, nullptr);
        pthread_mutex_init(mut + i, nullptr);
        pthread_create(slave_threads + i, nullptr, slave_main, (void*)(wc+i));
    }
#else
    slave_threads = (pthread_t *)emalloc(sizeof(pthread_t)*nrn_nthread);
#endif /* PERMANENT */
    if (!_interpreter_lock) {
        interpreter_locked = 0;
        _interpreter_lock = &interpreter_lock_;
        pthread_mutex_init(_interpreter_lock, nullptr);
    }
    if (!_nmodlmutex) {
        _nmodlmutex = &nmodlmutex_;
        pthread_mutex_init(_nmodlmutex, nullptr);
    }
    if (!_nrn_malloc_mutex) {
        _nrn_malloc_mutex = &nrn_malloc_mutex_;
        pthread_mutex_init(_nrn_malloc_mutex, nullptr);
    }
    nrn_thread_parallel_ = 1;
    }else{
        nrn_thread_parallel_ = 0;
    }
}

static void threads_free_pthread(){
    int i;
    if (slave_threads) {
#if PERMANENT
        wait_for_workers();
        for (i=1; i < nrn_nthread; ++i) {
            pthread_mutex_lock(mut + i);
            wc[i].flag = -1;
            pthread_cond_signal(cond + i);
            pthread_mutex_unlock(mut + i);
            pthread_join(slave_threads[i], nullptr);
            pthread_cond_destroy(cond + i);
            pthread_mutex_destroy(mut + i);
        }
        free((char*)slave_threads);
        free((char*)cond);
        free((char*)mut);
        free((char*)wc);
        slave_threads = (pthread_t*)0;
        cond = (pthread_cond_t*)0;
        mut = (pthread_mutex_t*)0;
        wc = (slave_conf_t*)0;
#else
        free((char*)slave_threads);
        slave_threads = (pthread_t*)0;
#endif /*PERMANENT*/
    }
    if (_interpreter_lock) {
        pthread_mutex_destroy(_interpreter_lock);
        _interpreter_lock = (pthread_mutex_t*)0;
        interpreter_locked = 0;
    }
    if (_nmodlmutex) {
        pthread_mutex_destroy(_nmodlmutex);
        _nmodlmutex = (pthread_mutex_t*)0;
    }
    if (_nrn_malloc_mutex) {
        pthread_mutex_destroy(_nrn_malloc_mutex);
        _nrn_malloc_mutex = (pthread_mutex_t*)0;
    }
    nrn_thread_parallel_ = 0;
}

#else /* USE_PTHREAD */

extern "C" void nrn_malloc_lock() {}
extern "C" void nrn_malloc_unlock() {}

static void threads_create_pthread(){
    nrn_thread_parallel_ = 0;
}
static void threads_free_pthread(){
    nrn_thread_parallel_ = 0;
}
#endif /* !USE_PTHREAD */

void nrn_thread_error(const char* s) {
    if (nrn_nthread != 1) {
        hoc_execerror(s, (char*)0);
    }
}

void nrn_thread_stat() {
#if BENCHMARKING
    FILE* f;
    long i, j, n;
    char buf[50];
    sprintf(buf, "bench.%d.dat", nrnmpi_myid);
    f = fopen(buf, "w");
#if 1
    n = (t_[0] - t1_[0]);
    for (i=1; i < nrn_nthread; ++i) {
        t_[i] = t1_[i] + n;
        t_[i+nrn_nthread] = t1_[i+nrn_nthread] + n;
    }
#endif
    n = 0;
    for (i=0; i < BS; ++i) {
        n += t_[i] - t1_[i];    
    }
    fprintf(f, "%ld\n", n);
    n = 0;
    for (j=0; j < BS; ++j) {
        n = t_[j] - t1_[j];
        for (i=0; i < n; ++i) {
            fprintf(f, "%ld %d\n", t1_[j][i], j*nrnmpi_numprocs + nrnmpi_myid);
        }
    }
    fclose(f);
#endif /*BENCHMARKING*/
}


void nrn_threads_create(int n, int parallel) {
    int i, j;
    NrnThread *nt;
    if (nrn_nthread != n) {
/*printf("sizeof(NrnThread)=%d   sizeof(Memb_list)=%d\n", sizeof(NrnThread), sizeof(Memb_list));*/
        threads_free_pthread();
        nrn_threads_free();
        for (i = 0; i < nrn_nthread; ++i) {
            nt = nrn_threads + i;
            if (nt->userpart) {
                hoc_obj_unref(nt->userpart);
            }
        }
        free((char *) nrn_threads);
#if BENCHMARKING
#endif
        nrn_threads = (NrnThread *) 0;
        nrn_nthread = n;
        if (n > 0) {
            CACHELINE_ALLOC(nrn_threads, NrnThread, n);
#if BENCHMARKING
            for (i=0; i < BS; ++i) {
                t_[i] = t1_[i];
            }
#endif
            for (i = 0; i < n; ++i) {
                nt = nrn_threads + i;
                nt->_t = 0.;
                nt->_dt = -1e9;
                nt->id = i;
                nt->_stop_stepping = 0;
                nt->tml = (NrnThreadMembList *) 0;
                nt->_ml_list = NULL;
                nt->roots = (hoc_List *) 0;
                nt->userpart = 0;
                nt->ncell = 0;
                nt->end = 0;
                for (j = 0; j < BEFORE_AFTER_SIZE; ++j) {
                    nt->tbl[j] = (NrnThreadBAList *) 0;
                }
                nt->_actual_rhs = 0;
                nt->_actual_d = 0;
                nt->_actual_a = 0;
                nt->_actual_b = 0;
                nt->_actual_v = 0;
                nt->_actual_area = 0;
                nt->_v_parent_index = 0;
                nt->_v_node = 0;
                nt->_v_parent = 0;
                nt->_ecell_memb_list = 0;
                nt->_ecell_child_cnt = 0;
                nt->_ecell_children = NULL;
                nt->_sp13mat = 0;
                nt->_ctime = 0.0;
                nt->_vcv = 0;
                nt->_nrn_fast_imem = 0;
            }
        }
        v_structure_change = 1;
        diam_changed = 1;
    }
    if (nrn_thread_parallel_ != parallel) {
        threads_free_pthread();
        if (parallel) {
            threads_create_pthread();
        }
    }
    /*printf("nrn_threads_create %d %d\n", nrn_nthread, nrn_thread_parallel_);*/
}

/*
Avoid invalidating pointers to i_membrane_ unless the number of compartments
in a thread has changed.
*/
static int fast_imem_nthread_ = 0;
static int* fast_imem_size_ = NULL;
static _nrn_Fast_Imem* fast_imem_;

static void fast_imem_free() {
    int i;
    for (i = 0; i < nrn_nthread; ++i) {
        nrn_threads[i]._nrn_fast_imem = NULL;
    }
    for (i = 0; i < fast_imem_nthread_; ++i) {
        if (fast_imem_size_[i] > 0) {
            free(fast_imem_[i]._nrn_sav_rhs);
            free(fast_imem_[i]._nrn_sav_d);
        }
    }
    if (fast_imem_nthread_) {
        free(fast_imem_size_);
        free(fast_imem_);
        fast_imem_nthread_ = 0;
        fast_imem_size_ = NULL;
        fast_imem_ = NULL;
    }
}

static void fast_imem_alloc() {
    int i;
    if (fast_imem_nthread_ != nrn_nthread) {
        fast_imem_free();
        fast_imem_nthread_ = nrn_nthread;
        fast_imem_size_ = static_cast<int*>(ecalloc(nrn_nthread, sizeof(int)));
        fast_imem_ = (_nrn_Fast_Imem*)ecalloc(nrn_nthread, sizeof(_nrn_Fast_Imem));
    }
    for (i=0; i < nrn_nthread; ++i) {
        NrnThread* nt = nrn_threads + i;
        int n = nt->end;
        _nrn_Fast_Imem* fi = fast_imem_ + i;
        if (n != fast_imem_size_[i]) {
            if (fast_imem_size_[i] > 0) {
                free(fi->_nrn_sav_rhs);
                free(fi->_nrn_sav_d);
            }
            if (n > 0) {
                CACHELINE_CALLOC(fi->_nrn_sav_rhs, double, n);
                CACHELINE_CALLOC(fi->_nrn_sav_d, double, n);
            }
            fast_imem_size_[i] = n;
        }
    }
}

void nrn_fast_imem_alloc() {
    if (nrn_use_fast_imem) {
        int i;
        fast_imem_alloc();
        for (i=0; i < nrn_nthread; ++i) {
          nrn_threads[i]._nrn_fast_imem = fast_imem_ + i;
        }
    }else{
        fast_imem_free();
    }
}

void nrn_threads_free() {
    int it, i;
    for (it = 0; it < nrn_nthread; ++it) {
        NrnThread* nt = nrn_threads + it;
        NrnThreadMembList* tml, *tml2;
        for (tml = nt->tml; tml; tml = tml2) {
            Memb_list* ml = tml->ml;
            tml2 = tml->next;
            free((char*)ml->nodelist);
            free((char*)ml->nodeindices);
            if (memb_func[tml->index].hoc_mech) {
                free((char*)ml->prop);
            }else{
                free((char*)ml->data);
                free((char*)ml->pdata);
            }
            if (ml->_thread) {
                if (memb_func[tml->index].thread_cleanup_) {
    (*memb_func[tml->index].thread_cleanup_)(ml->_thread);
                }
                free((char*)ml->_thread);
            }
            free((char*)ml);
            free((char*)tml);
        }
        if (nt->_ml_list) { free((char*)nt->_ml_list); nt->_ml_list = NULL; }
        for (i=0; i < BEFORE_AFTER_SIZE; ++i) {
            NrnThreadBAList* tbl, *tbl2;
            for (tbl = nt->tbl[i]; tbl; tbl = tbl2) {
                tbl2 = tbl->next;
                free((char*)tbl);
            }
            nt->tbl[i] = (NrnThreadBAList*)0;
        }
        nt->tml = (NrnThreadMembList*)0;
        if (nt->userpart == 0 && nt->roots) {
            hoc_l_freelist(&nt->roots);
            nt->ncell = 0;
        }
        if (nt->_actual_rhs) {free((char*)nt->_actual_rhs); nt->_actual_rhs = 0;}
        if (nt->_actual_d) {free((char*)nt->_actual_d); nt->_actual_d = 0;}
        if (nt->_actual_a) {free((char*)nt->_actual_a); nt->_actual_a = 0;}
        if (nt->_actual_b) {free((char*)nt->_actual_b); nt->_actual_b = 0;}
        if (nt->_v_parent_index) {free((char*)nt->_v_parent_index); nt->_v_parent_index = 0;}
        if (nt->_v_node) {free((char*)nt->_v_node); nt->_v_node = 0;}
        if (nt->_v_parent) {free((char*)nt->_v_parent); nt->_v_parent = 0;}
        nt->_ecell_memb_list = 0;
        if (nt->_ecell_children) {
            nt->_ecell_child_cnt = 0;
            free(nt->_ecell_children);
            nt->_ecell_children = NULL;
        }
        if (nt->_sp13mat) {
            spDestroy(nt->_sp13mat);
            nt->_sp13mat = 0;
        }
        nt->_nrn_fast_imem = NULL;
        /* following freed by nrn_recalc_node_ptrs */
        nrn_old_thread_save();
        nt->_actual_v = 0;      
        nt->_actual_area = 0;
        nt->end = 0;
        nt->ncell = 0;
        nt->_vcv = 0;
    }
}

/* be careful to make the tml list in proper memb_order. This is important */
/* for correct finitialize where mechanisms that write concentrations must be */
/* after ions and before mechanisms that read concentrations. */

static void thread_memblist_setup(NrnThread* _nt, int* mlcnt, void** vmap) {
    int i, ii;
    Node* nd;
    Prop* p;
    NrnThreadMembList* tml, **ptml;
    Memb_list** mlmap = (Memb_list**)vmap;  
    BAMech** bamap = (BAMech**)vmap;
#if 0
printf("thread_memblist_setup %lx v_node_count=%d ncell=%d end=%d\n", (long)nth, v_node_count, nth->ncell, nth->end);
#endif
    for (i=0; i < n_memb_func; ++i) {
        mlcnt[i] = 0;
    }

    /* count */
    for (i = 0; i < _nt->end; ++i) {
        nd = _nt->_v_node[i];
        for (p = nd->prop; p; p = p->next) {
            if (memb_func[p->type].current
                || memb_func[p->type].state
                || memb_func[p->type].initialize
            ) {
                ++mlcnt[p->type];
            }
        }
    }
    /* allocate */
    ptml = &_nt->tml;
    for (ii = 0; ii < n_memb_func; ++ii) {
        i = memb_order_[ii];
        if (mlcnt[i]) {
            if (_nt->id > 0 && memb_func[i].vectorized == 0) {
hoc_execerror(memb_func[i].sym->name, "is not thread safe");
            }
/*printf("thread_memblist_setup %lx type=%d cnt=%d\n", (long)nth, i, mlcnt[i]);*/
            CACHELINE_ALLOC(tml, NrnThreadMembList, 1);
            tml->index = i;
            tml->next = (NrnThreadMembList*)0;
            *ptml = tml;
            ptml = &tml->next;
            CACHELINE_ALLOC(tml->ml, Memb_list, 1);
            if (i == EXTRACELL) { _nt->_ecell_memb_list = tml->ml; }
            mlmap[i] = tml->ml;
            CACHELINE_ALLOC(tml->ml->nodelist, Node*, mlcnt[i]);
            CACHELINE_ALLOC(tml->ml->nodeindices, int, mlcnt[i]);
            if (memb_func[i].hoc_mech) {
                tml->ml->prop = (Prop**)emalloc(mlcnt[i]*sizeof(Prop*));
            }else{
                CACHELINE_ALLOC(tml->ml->data, double*, mlcnt[i]);
                CACHELINE_ALLOC(tml->ml->pdata, Datum*, mlcnt[i]);
            }
            tml->ml->_thread = (Datum*)0;
            if (memb_func[i].thread_size_) {
                tml->ml->_thread = (Datum*)ecalloc(memb_func[i].thread_size_, sizeof(Datum));
                if (memb_func[tml->index].thread_mem_init_) {
    (*memb_func[tml->index].thread_mem_init_)(tml->ml->_thread);
                }
            }
            tml->ml->nodecount = 0; /* counted again below */
        }
    }
    CACHELINE_CALLOC(_nt->_ml_list, Memb_list*, n_memb_func);
    for (tml = _nt->tml; tml; tml = tml->next) {
        _nt->_ml_list[tml->index] = tml->ml;
    }

    /* fill */
    for (i = 0; i < _nt->end; ++i) {
        nd = _nt->_v_node[i];
        for (p = nd->prop; p; p = p->next) {
            if (memb_func[p->type].current
                || memb_func[p->type].state
                || memb_func[p->type].initialize
            ) {
                Memb_list* ml = mlmap[p->type];
                ml->nodelist[ml->nodecount] = nd;
                ml->nodeindices[ml->nodecount] = nd->v_node_index;
                if (memb_func[p->type].hoc_mech) {
                    ml->prop[ml->nodecount] = p;
                }else{
                    ml->data[ml->nodecount] = p->param;
                    ml->pdata[ml->nodecount] = p->dparam;
                }
                ++ml->nodecount;
            }
        }
    }
    /* count and store any Node* with the property
       nd->extnode == NULL && nd->pnd != NULL && nd->pnd->extcell != NULL
    */
    if (_nt->_ecell_memb_list) {
        Node* pnd;
        int cnt=0;
        for (i = 0; i < _nt->end; ++i) {
            nd = _nt->_v_node[i];
            pnd = _nt->_v_parent[i];
            if (nd->extnode == NULL && pnd && pnd->extnode) {
                ++cnt;
            }
        }
        if (cnt) {
            Node** p;
            CACHELINE_ALLOC(_nt->_ecell_children, Node*, cnt);
            _nt->_ecell_child_cnt = cnt;
            p = _nt->_ecell_children;
            cnt = 0;
            for (i = 0; i < _nt->end; ++i) {
                nd = _nt->_v_node[i];
                pnd = _nt->_v_parent[i];
                if (nd->extnode == NULL && pnd && pnd->extnode) {
                    p[cnt++] = nd;
                }
            }
        }
    }
#if 0
    for (i=0; i < n_memb_func; ++i) {
        if (mlcnt[i]) {assert(mlcnt[i] = mlmap[i]->nodecount);}
    }
    for (tml = _nt->tml; tml; tml = tml->next) {
        assert(mlcnt[tml->index] == tml->ml->nodecount);
    }
#endif
    /* fill the ba lists */
    /* need map from ml type to BA type. Reuse vmap */
    for (i=0; i < BEFORE_AFTER_SIZE; ++i) {
        BAMech* bam;
        NrnThreadBAList* tbl, **ptbl;
        for (ii=0; ii < n_memb_func; ++ii) {
            bamap[ii] = (BAMech*)0;
        }
        for (bam = bamech_[i]; bam; bam = bam->next) {
            bamap[bam->type] = bam;
        }
        /* unnecessary but keep in order anyway */
        ptbl = _nt->tbl + i;
        for (tml = _nt->tml; tml; tml = tml->next) {
            if (bamap[tml->index]) {
                Memb_list* ml = tml->ml;
                tbl = (NrnThreadBAList*)emalloc(sizeof(NrnThreadBAList));
                tbl->next = (NrnThreadBAList*)0;
                tbl->bam = bamap[tml->index];
                tbl->ml = ml;
                *ptbl = tbl;
                ptbl = &(tbl->next);
            }
        }
    }
    /* fill in the Point_process._vnt value. */
    /* artificial cells done in v_setup_vectors() */
    for (tml = _nt->tml; tml; tml = tml->next) if (memb_func[tml->index].is_point) {
        for (i=0; i < tml->ml->nodecount; ++i) {
            Point_process* pnt = (Point_process*)tml->ml->pdata[i][1]._pvoid;
            pnt->_vnt = (void*)_nt;
        }
    }
}

static void nrn_thread_memblist_setup() {
    int it, *mlcnt;
    void** vmap;
    mlcnt = (int*)emalloc(n_memb_func*sizeof(int));
    vmap = (void**)emalloc(n_memb_func*sizeof(void*));
    for (it=0; it < nrn_nthread; ++it) {
        thread_memblist_setup(nrn_threads + it, mlcnt, vmap);
    }
    nrn_fast_imem_alloc();
    free((char*)vmap);
    free((char*)mlcnt);
    nrn_mk_table_check();
    if (nrn_mk_transfer_thread_data_) { (*nrn_mk_transfer_thread_data_)(); }
}

/* secorder needs to correspond to cells in NrnThread with roots */
/* at the beginning of each thread region */
/* this differs from original secorder where all roots are at the beginning */
/* in passing, also set start and end indices. */

static void reorder_secorder() {
    NrnThread* _nt;
    Section* sec, *ch;
    Node* nd;
    hoc_Item* qsec;
    hoc_List* sl;
    int order, isec, i, j, inode;
    /* count and allocate */
    ForAllSections(sec)
        sec->order = -1;
    }
    order = 0;
    FOR_THREADS(_nt) {
        /* roots of this thread */
        sl = _nt->roots;
        inode = 0;
        ITERATE(qsec, sl) {
            sec = hocSEC(qsec);
            assert(sec->order == -1);
            secorder[order] = sec;
            sec->order = order;
            ++order;
            nd = sec->parentnode;
            nd->_nt = _nt;
            inode += 1;
        }
        /* all children of what is already in secorder */
        for (isec = order - _nt->ncell; isec < order; ++isec) {
            sec = secorder[isec];
            /* to make it easy to fill in PreSyn.nt_*/
            sec->prop->dparam[9]._pvoid = (void*)_nt;
            for (j = 0; j < sec->nnode; ++j) {
                nd = sec->pnode[j];
                nd->_nt = _nt;
                inode += 1;
            }
            for (ch = sec->child; ch; ch = ch->sibling) {
                assert(ch->order == -1);
                secorder[order] = ch;
                ch->order = order;
                ++order;
            }
        }
        _nt->end = inode;
        CACHELINE_CALLOC(_nt->_actual_rhs, double, inode);
        CACHELINE_CALLOC(_nt->_actual_d, double, inode);
        CACHELINE_CALLOC(_nt->_actual_a, double, inode);
        CACHELINE_CALLOC(_nt->_actual_b, double, inode);
        CACHELINE_CALLOC(_nt->_v_node, Node*, inode);
        CACHELINE_CALLOC(_nt->_v_parent, Node*, inode);
        CACHELINE_CALLOC(_nt->_v_parent_index, int, inode);
    }
    /* do it again and fill _v_node and _v_parent */
    /* index each cell section in relative order. Do offset later */
    ForAllSections(sec)
        sec->order = -1;
    }
    order = 0;
    FOR_THREADS(_nt) {
        /* roots of this thread */
        sl = _nt->roots;
        inode = 0;
        ITERATE(qsec, sl) {
            sec = hocSEC(qsec);
            assert(sec->order == -1);
            secorder[order] = sec;
            sec->order = order;
            ++order;
            nd = sec->parentnode;
            nd->_nt = _nt;
            _nt->_v_node[inode] = nd;
            _nt->_v_parent[inode] = (Node*)0;
            _nt->_v_node[inode]->v_node_index = inode;
            inode += 1;
        }
        /* all children of what is already in secorder */
        for (isec = order - _nt->ncell; isec < order; ++isec) {
            sec = secorder[isec];
            /* to make it easy to fill in PreSyn.nt_*/
            sec->prop->dparam[9]._pvoid = (void*)_nt;
            for (j = 0; j < sec->nnode; ++j) {
                nd = sec->pnode[j];
                nd->_nt = _nt;
                _nt->_v_node[inode] = nd;
                if (j) {
                    _nt->_v_parent[inode] = sec->pnode[j - 1];
                }else{
                    _nt->_v_parent[inode] = sec->parentnode;
                }
                _nt->_v_node[inode]->v_node_index = inode;
                inode += 1;
            }
            for (ch = sec->child; ch; ch = ch->sibling) {
                assert(ch->order == -1);
                secorder[order] = ch;
                ch->order = order;
                ++order;
            }
        }
        _nt->end = inode;
    }
    assert(order == section_count);
    /*assert(inode == v_node_count);*/
    /* not missing any */
    ForAllSections(sec) assert(sec->order != -1); }

    /* here is where multisplit reorders the nodes. Afterwards
      in either case, we can then point to v, d, rhs in proper
      node order
    */
    FOR_THREADS(_nt) for (inode=0; inode < _nt->end; ++inode) {
        _nt->_v_node[inode]->_classical_parent = _nt->_v_parent[inode];
    }
    if (nrn_multisplit_setup_) {
        /* classical order abandoned */
        (*nrn_multisplit_setup_)();
    }
    /* make the Nodes point to the proper d, rhs */
    FOR_THREADS(_nt) {
        for (j = 0; j < _nt->end; ++j) {
            Node* nd = _nt->_v_node[j];
            nd->_d = _nt->_actual_d + j;
            nd->_rhs = _nt->_actual_rhs + j;
        }
    }
    /* because the d,rhs changed, if multisplit is used we need to update
      the reduced tree gather/scatter pointers
    */
    if (nrn_multisplit_setup_) {
        nrn_multisplit_ptr_update();
    }
}


void nrn_mk_table_check() {
    int i, id, index;
    int* ix;
    if (table_check_) {
        free((void*)table_check_);
        table_check_ = (Datum*)0;
    }
    ix = (int*)emalloc(n_memb_func*sizeof(int));
    for (i=0; i < n_memb_func; ++i) {
        ix[i] = -1;
    }
    table_check_cnt_ = 0;
    for (id=0; id < nrn_nthread; ++id) {
        NrnThread* nt = nrn_threads + id;
        NrnThreadMembList* tml;
        for (tml = nt->tml; tml; tml = tml->next) {
            index = tml->index;
            if (memb_func[index].thread_table_check_ && ix[index] == -1) {
                ix[index] = id;
                table_check_cnt_ += 2;
            }
        }
    }
    if (table_check_cnt_) {
        table_check_ = (Datum*)emalloc(table_check_cnt_*sizeof(Datum));
    }
    i=0;
    for (id=0; id < nrn_nthread; ++id) {
        NrnThread* nt = nrn_threads + id;
        NrnThreadMembList* tml;
        for (tml = nt->tml; tml; tml = tml->next) {
            index = tml->index;
            if (memb_func[index].thread_table_check_ && ix[index] == id) {
                table_check_[i++].i = id;
                table_check_[i++]._pvoid = (void*)tml;
            }
        }
    }
    free((void*)ix);
}

void nrn_thread_table_check() {
    int i;
    for (i=0; i < table_check_cnt_; i += 2) {
        NrnThread* nt = nrn_threads + table_check_[i].i;
        NrnThreadMembList* tml = (NrnThreadMembList*)table_check_[i+1]._pvoid;
        Memb_list* ml = tml->ml;
        (*memb_func[tml->index].thread_table_check_)(
            ml->data[0], ml->pdata[0], ml->_thread, nt, tml->index
        );
    }
}

/* if it is possible for more than one thread to get into the
   interpreter, lock it. */
void nrn_hoc_lock() {
#if USE_PTHREAD
    if (nrn_inthread_) {
        pthread_mutex_lock(_interpreter_lock);
        interpreter_locked = 1;
    }
#endif
}
void nrn_hoc_unlock() {
#if USE_PTHREAD
    if (interpreter_locked) {
        interpreter_locked = 0;
        pthread_mutex_unlock(_interpreter_lock);
    }
#endif
}

void nrn_multithread_job(void*(*job)(NrnThread*)) {
    int i;
#if USE_PTHREAD
    BENCHDECLARE
    if (nrn_thread_parallel_) {
        nrn_inthread_ = 1;
        for (i=1; i < nrn_nthread; ++i) {
            send_job_to_slave(i, job);
        }
        BENCHBEGIN(0)
        (*job)(nrn_threads);
        BENCHADD(nrn_nthread)
        WAIT();
        nrn_inthread_ = 0;
    }else{ /* sequential */
#else
    {
#endif
        for (i=1; i < nrn_nthread; ++i) {
            BENCHBEGIN(i)
            (*job)(nrn_threads + i);
            BENCHADD(i+nrn_nthread)
        }
        BENCHBEGIN(0)
        (*job)(nrn_threads);
        BENCHADD(nrn_nthread)
    }
}


void nrn_onethread_job(int i, void*(*job)(NrnThread*)) {
    BENCHDECLARE
    assert(i >= 0 && i < nrn_nthread);
#if USE_PTHREAD
    if (nrn_thread_parallel_) {
        if (i > 0) {
            send_job_to_slave(i, job);
            WAIT();
        }else{
            BENCHBEGIN(0)
            (*job)(nrn_threads);
            BENCHADD(nrn_nthread)
        }
    }else{
#else
    {
#endif
        (*job)(nrn_threads + i);
    }
}

void nrn_wait_for_threads() {
#if USE_PTHREAD
    if (nrn_thread_parallel_) {
        wait_for_workers();
    }
#endif
}

void nrn_thread_partition(int it, Object* sl) {
    NrnThread* nt;
    assert(it >= 0 && it < nrn_nthread);
    nt = nrn_threads + it;
    if (nt->userpart == nullptr && nt->roots) {
        hoc_l_freelist(&nt->roots);
    }
    if (sl) {
        hoc_obj_ref(sl);
    }
    if (nt->userpart) {
        hoc_obj_unref(nt->userpart);
        nt->userpart = nullptr;
        nt->roots = (hoc_List*)0;
    }
    if (sl) {
        nt->userpart = sl; /* already reffed above */
        nt->roots = (hoc_List*)sl->u.this_pointer;
    }
    v_structure_change = 1;
}

int nrn_user_partition() {
    int i, it, b, n;
    hoc_Item* qsec;
    hoc_List* sl;
    char buf[256];
    Section* sec;
    NrnThread* nt;
    /* all one or all the other*/
    b = (nrn_threads[0].userpart != nullptr);
    for (it=1; it < nrn_nthread; ++it) {
        if ((nrn_threads[it].userpart != nullptr) != b) {
hoc_execerror("some threads have a user defined partition", "and some do not");
        }
    }
    if (!b) { return 0; }
    
    /* discard partition if any section mentioned has been deleted. The
        model has changed */
    FOR_THREADS(nt) {
        sl = nt->roots;
        ITERATE(qsec, sl) {
            sec = hocSEC(qsec);
            if (!sec->prop) {
                for (i=0; i < nrn_nthread; ++i) {
                    nrn_thread_partition(i, nullptr);
                }
                return 0;
            }
        }
    }
    
    ForAllSections(sec)
        sec->volatile_mark = 0;
    }
    /* fill in ncell and verify consistency */
    n = 0;
    for (it=0; it < nrn_nthread; ++it) {
        nt = nrn_threads + it;
        sl = nt->roots;
        nt->ncell = 0;
        ITERATE(qsec, sl) {
            sec = hocSEC(qsec);
            ++nt->ncell;
            ++n;
            if (sec->parentsec) {
sprintf(buf, "in thread partition %d is not a root section", it);
                hoc_execerror(secname(sec), buf);
            }
            if (sec->volatile_mark) {
sprintf(buf, "appeared again in partition %d", it);
                hoc_execerror(secname(sec), buf);
            }
            sec->volatile_mark = 1;
        }
    }
    if (n  != nrn_global_ncell) {
sprintf(buf, "The total number of cells, %d, is different than the number of user partition cells, %d\n", nrn_global_ncell, n);
        hoc_execerror(buf, (char*)0);
    }
    return 1;
}

void nrn_use_busywait(int b) {
#if USE_PTHREAD
    if (allow_busywait_ && nrn_thread_parallel_) {
        if (b == 0 && busywait_main_ == 1) {
            busywait_ = 0;
            nrn_multithread_job(nulljob);
            busywait_main_ = 0;
        }else if (b == 1 && busywait_main_ == 0) {
            busywait_main_ = 1;
            wait_for_workers();
            busywait_ = 1;
            nrn_multithread_job(nulljob);
        }
    }else{
        if (busywait_main_ == 1) {
            busywait_ = 0;
            nrn_multithread_job(nulljob);
            busywait_main_ = 0;
        }
    }
#endif
}

int nrn_allow_busywait(int b) {
    int old = allow_busywait_;
    allow_busywait_ = b;
    return old;
}

#if USE_PTHREAD
static long waste_;
static void* waste(void* v) {
    size_t i, j, n;
    n = (size_t)v;
    j = 0;
    for (i=0; i < n; ++i) {
        j += i;
    }
    /* hoping it is  not optimized away */
    waste_ = j;
    return nullptr;
}

#define _nt_ 32
static double trial(int ip) {
    int i;
    double t;
    pthread_t* th;
    th = (pthread_t*)ecalloc(ip, sizeof(pthread_t));
    t = nrn_timeus();
    for (i=0; i < ip; ++i) {
        pthread_create(th + i, nullptr, waste, (void*)100000000);
    }
    for (i=0; i < ip; ++i) {
        pthread_join(th[i], nullptr);
    }
    t = nrn_timeus() - t;
    free((char*)th);
    return t;   
}
#endif

int nrn_how_many_processors() {
#if USE_PTHREAD
    int i, ip;
    double t1, t2;
    printf("nthread walltime (count to 1e8 on each thread)\n");
    t1 = trial(1);
    printf("%4d\t %g\n", 1, t1);
    for (ip = 2; ip <= _nt_; ip *= 2) {
        t2 = trial(ip);
        printf("%4d\t %g\n", ip, t2);
        if (t2 > 1.3*t1) {
            return ip/2;
        }
    }
    return _nt_;
#else
    return 1;
#endif
}