Yield results asynchronously in python using multiprocessing or twisted -


I have an embarrassing parallel application, where the order of results does not matter.

I have a function and a list of 1000 arguments on which it works.

  def _process_parallel (function, args_list, args_dict = {});   

I have written a few multi codes for this parallelize. ..: num_tasks = lane (args_list) num_tasks_returned_ptr = [0] def _callback (result): num_tasks_returned_ptr [0] + = 1 # All jobs must be executed asynconously apply_results = [__POOL __ send args to apply_async (function, arg , Args_dict, _callback). Args_list] Wait for # until all tasks are processed num_tasks_returned_ptr [0]

I think the memory footprint is too high, the results of the function are not left at present until all the results are processed.

What I would like to do instead, there is something where the results are not stored after the execution. Something like this:

  def _process_parallel (function, args_list, args_dict = {}): # Send all jobs to the result in somepackage.apply_async to be executed asynconously (function, arg, args_dict, _callback): yield result  

I can not seem to find a way in multiplying I've heard good things about twisted, but I'm not sure this is simple After this for work

make a dragon generator You know the results regarding the area to calculate the asynchronous methods and they come like that causes them?


Comments

Popular posts from this blog

Editing Python Class in Shell and SQLAlchemy -

uislider - In a MATLAB GUI, how does one implement a continuously varying slider from a GUIDE created .m file? -

import - Python ImportError: No module named wmi -