blis/frame/base/bli_threading.c

/*

   BLIS    
   An object-based framework for developing high-performance BLAS-like
   libraries.

   Copyright (C) 2014, The University of Texas at Austin

   Redistribution and use in source and binary forms, with or without
   modification, are permitted provided that the following conditions are
   met:
    - Redistributions of source code must retain the above copyright
      notice, this list of conditions and the following disclaimer.
    - Redistributions in binary form must reproduce the above copyright
      notice, this list of conditions and the following disclaimer in the
      documentation and/or other materials provided with the distribution.
    - Neither the name of The University of Texas at Austin nor the names
      of its contributors may be used to endorse or promote products
      derived from this software without specific prior written permission.

   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
   HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

*/

#include "blis.h"

#ifdef BLIS_TREE_BARRIER
void bli_free_barrier_tree( barrier_t* barrier )
{
    if( barrier == NULL )
        return;
    barrier->count--;
    if( barrier->count == 0 )
    {
        bli_free_barrier_tree( barrier->dad );
        bli_free( barrier );
    }
    return;
}
barrier_t* bli_create_tree_barrier(int num_threads, int arity, barrier_t** leaves, int leaf_index)
{
    barrier_t* me = (barrier_t*) malloc(sizeof(barrier_t));

    me->dad = NULL;
    me->signal = 0;

    // Base Case
    if( num_threads <= arity ) {
       //Now must be registered as a leaf
       for(int i = 0; i < num_threads; i++)
       {
           leaves[leaf_index + i] = me;
       }
       me->count = num_threads;
       me->arity = num_threads;
    }
    else {
        // Otherwise this node has children
        int threads_per_kid = num_threads / arity;
        int defecit = num_threads - threads_per_kid * arity;

        for(int i = 0; i < arity; i++){
            int threads_this_kid = threads_per_kid;
            if(i < defecit) threads_this_kid++;
            
            barrier_t* kid = bli_create_tree_barrier(threads_this_kid, arity, leaves, leaf_index);
            kid->dad = me;

            leaf_index += threads_this_kid;
        }  
        me->count = arity;
        me->arity = arity;
    }  

    return me;
}

void bli_cleanup_communicator( thread_comm_t* communicator )
{
    if( communicator == NULL ) return;
    for( dim_t i = 0; i < communicator->n_threads; i++)
    {
       bli_free_barrier_tree( communicator->barriers[i] );
    }
    bli_free( communicator->barriers );
}
void bli_setup_communicator( thread_comm_t* communicator, dim_t n_threads)
{
    if( communicator == NULL ) return;
    communicator->sent_object = NULL;
    communicator->n_threads = n_threads;
    communicator->barriers = ( barrier_t** ) bli_malloc( sizeof( barrier_t* ) * n_threads );
    bli_create_tree_barrier( n_threads, BLIS_TREE_BARRIER_ARITY, communicator->barriers, 0 );
}

void tree_barrier( barrier_t* barack )
{
#ifdef BLIS_ENABLE_OPENMP
    int my_signal = barack->signal;
    int my_count;

    _Pragma("omp atomic capture")
        my_count = barack->count--;

    if( my_count == 1 ) {
        if(  barack->dad != NULL ) {
            tree_barrier( barack->dad );
        }
        barack->count = barack->arity;
        barack->signal = !barack->signal;
    }
    else {
        volatile int* listener = &barack->signal;
        while( *listener == my_signal ) {}
    }
#else
    return
#endif
}

void bli_barrier( thread_comm_t* comm, dim_t t_id )
{
    tree_barrier( comm->barriers[t_id] );
}

#else

void bli_cleanup_communicator( thread_comm_t* communicator )
{
    if( communicator == NULL ) return;
}
void bli_setup_communicator( thread_comm_t* communicator, dim_t n_threads)
{
    if( communicator == NULL ) return;
    communicator->sent_object = NULL;
    communicator->n_threads = n_threads;
    communicator->barrier_sense = 0;
    communicator->barrier_threads_arrived = 0;
}
//barrier routine taken from art of multicore programming or something
void bli_barrier( thread_comm_t* communicator, dim_t t_id )
{
#ifdef BLIS_ENABLE_OPENMP
    if(communicator == NULL || communicator->n_threads == 1)
        return;
    bool_t my_sense = communicator->barrier_sense;
    dim_t my_threads_arrived;

    _Pragma("omp atomic capture")
        my_threads_arrived = ++(communicator->barrier_threads_arrived);

    if( my_threads_arrived == communicator->n_threads ) {
        communicator->barrier_threads_arrived = 0;
        communicator->barrier_sense = !communicator->barrier_sense;
    }
    else {
        volatile bool_t* listener = &communicator->barrier_sense;
        while( *listener == my_sense ) {}
    }
#else
    return;
#endif
}
#endif

void bli_free_communicator( thread_comm_t* communicator )
{
    if( communicator == NULL ) return;
    bli_cleanup_communicator( communicator );
    bli_free( communicator );
}

thread_comm_t* bli_create_communicator( dim_t n_threads )
{
    thread_comm_t* comm = (thread_comm_t*) bli_malloc( sizeof(thread_comm_t) );
    bli_setup_communicator( comm, n_threads );
    return comm;
}

void* bli_broadcast_structure( thread_comm_t* communicator, dim_t id, void* to_send )
{   
    if( communicator == NULL || communicator->n_threads == 1 ) return to_send;

    if( id == 0 ) communicator->sent_object = to_send;

    bli_barrier( communicator, id );
    void * object = communicator->sent_object;
    bli_barrier( communicator, id );

    return object;
}

thrinfo_t* bli_create_thread_info( thread_comm_t* ocomm, dim_t ocomm_id, thread_comm_t* icomm, dim_t icomm_id,
                             dim_t n_way, dim_t work_id )
{

        thrinfo_t* thr = (thrinfo_t*) bli_malloc( sizeof(thrinfo_t) );
        bli_setup_thread_info( thr, ocomm, ocomm_id, icomm, icomm_id, n_way, work_id );
        return thr;
}

void bli_setup_thread_info( thrinfo_t* thr, thread_comm_t* ocomm, dim_t ocomm_id, thread_comm_t* icomm, dim_t icomm_id,
                             dim_t n_way, dim_t work_id )
{
        thr->ocomm = ocomm;
        thr->ocomm_id = ocomm_id;
        thr->icomm = icomm;
        thr->icomm_id = icomm_id;

        thr->n_way = n_way;
        thr->work_id = work_id;
}

void bli_get_range( void* thr, dim_t all_start, dim_t all_end, dim_t block_factor, dim_t* start, dim_t* end )
{
    thrinfo_t* thread = (thrinfo_t*) thr;
    dim_t n_way = thread->n_way;
    dim_t work_id = thread->work_id;

    dim_t size = all_end - all_start;
    dim_t n_pt = size / n_way;
    n_pt = (n_pt * n_way < size) ? n_pt + 1 : n_pt;
    n_pt = (n_pt % block_factor == 0) ? n_pt : n_pt + block_factor - (n_pt % block_factor); 
    *start = work_id * n_pt + all_start;
    *end   = bli_min( *start + n_pt, size + all_start );
}

void bli_get_range_weighted( void* thr, dim_t all_start, dim_t all_end, dim_t block_factor, bool_t forward, dim_t* start, dim_t* end)
{
    thrinfo_t* thread = (thrinfo_t*) thr;
    dim_t n_way = thread->n_way;
    dim_t work_id = thread->work_id;
    dim_t size = all_end - all_start;

    *start = 0;
    *end   = all_end - all_start;

    if( forward ) {
        dim_t curr_caucus = n_way - 1;
        dim_t len = 0;
        dim_t num = size*size / n_way; // 2xArea per thread?
        while(1){
            dim_t width = ceil(sqrt( len*len + num )) - len; // The width of the current caucus
            width = (width % block_factor == 0) ? width : width + block_factor - (width % block_factor);
            if( curr_caucus == work_id ) {
                *start = bli_max( 0 , *end - width ) + all_start;
                *end = *end + all_start;
                return;
            }
            else{
                *end -= width;
                len += width;
                curr_caucus--;
            }
        }
    }
    else{
        dim_t num = size*size / n_way;
        while(1){
            dim_t width = ceil(sqrt(*start * *start + num)) - *start;
            width = (width % block_factor == 0) ? width : width + block_factor - (width % block_factor);

            if( work_id == 0 ) {
                *start = *start + all_start;
                *end = bli_min( *start + width, all_end );
                return;
            }
            else{
                *start = *start + width;
            }
            work_id--;
        }
    }
}

void bli_level3_thread_decorator( dim_t n_threads, 
                                  level3_int_t func, 
                                  obj_t* alpha, 
                                  obj_t* a, 
                                  obj_t* b, 
                                  obj_t* beta, 
                                  obj_t* c, 
                                  void* cntl, 
                                  void** thread )
{
#ifdef BLIS_ENABLE_OPENMP
    _Pragma( "omp parallel num_threads(n_threads)" )
    {
        dim_t omp_id = omp_get_thread_num();

        func( alpha,
                  a,
                  b,
                  beta,
                  c,
                  cntl,
                  thread[omp_id] );
    }
#else
        func( alpha,
                  a,
                  b,
                  beta,
                  c,
                  cntl,
                  thread[0] );
#endif
}

dim_t bli_read_nway_from_env( char* env )
{
    dim_t number = 1;
    char* str = getenv( env );
    if( str != NULL )
    {   
        number = strtol( str, NULL, 10 );
    }   
    return number;
}