Using tasks for repeated actions RRS feed

  • Question

  • I have 50 Machine Learning agents. Every frame, they get some inputs and compute the neural network. Because every agent is independent, I would like to make every agent compute the network as a separate task.

    If I were to create a task for every agent, each frame, it will make my program slower. I tried to group my agents into 2 tasks (25 and 25), but it was still an overhead.

    The way I see it, is to create a task at agent construction. That task would compute my network. So every frame now, I would restart the same task, or something similar. In this way, I think I will avoid that extra overhead at the Task creation.

    I have read some articles on this topic, and I found out I can't reuse a task. So, what workaround could work?

    Basically, I have a repeated action on 50 agents, that is run every frame, for about a minute, and it would be a waste not to parallelize them.

    I am still new to multithreading and tasks, so I am relying on your help.

    Friday, June 8, 2018 8:33 PM