Mcore_handler

class nrv.backend.Mcore_handler(*args, **kwargs)

Class to handle parallel processing (cores, no threads) in NRV2

Methods

Mcore_handler.__init__(Flag)	Instantiation of Mcore
Mcore_handler.do_master_only_work()	Check if the process is alone or it is the master to perform non splitable job
Mcore_handler.gather_jobs(partial_result)	Gather the jobs performed by all instances to the master
Mcore_handler.gather_jobs_as_array(...)	Gather the jobs performed by all instances to the master
Mcore_handler.is_alone()	Check if the process is runing alone or if other instances have been launched.
Mcore_handler.is_master()	Check if the process is master or not whe parallel computing
Mcore_handler.master_broadcasts_array_to_all(var)	Broadcast an array to all instances of the process (share jobs performed by the master only)
Mcore_handler.master_broadcasts_to_all(var)	Broadcast an array to all instances of the process (share jobs performed by the master only)
Mcore_handler.recieve_data_from_slave()	Recieve data from anay source
Mcore_handler.recieve_potential_array_from_master()	Recieve potenatial data from the master as a numpy array
Mcore_handler.say_hello()	Display a sentence from each process on prompt.
Mcore_handler.send_back_array_to_dest(data, ...)	Send a numpy array to a slave
Mcore_handler.send_data_to_master(data)	Send a dictionary of data directly to the master.
Mcore_handler.send_synchronization_flag()	Blocking collective communication to force all process to synchronize to a specific line of code
Mcore_handler.split_job_from_arrays(len_arrays)	Split an array for parallel independant computing, by sharing independant sub-spaces of array index
Mcore_handler.split_job_from_arrays_to_slaves(...)	Split an array for parallel independant computing, by sharing independant sub-spaces of array index, the master gets a table of all jobs to do initialized to False
Mcore_handler.sum_jobs(partial_result)	Gather, sum and broadcast array for conservative results.