How to optimize my reinforcement learning training simulation on my M3Pro

Hello,

so I'm trying to run a RL training environment with Simulink. The Simulink environment is pretty intense, plus my agent is a custom agent that I created that has quite a alot of networks. I was just wondering what are the methods I can do to optimize training/simulation as much as possible because I don't want the simulation to run for like 24 hours and cause damage to my mac since Matlab simulation are pretty heavy. I read briefly about the the parallel computing tool box, but I don't know how that will interact with my customly created agent. Is there any setting I should change on my Mac as I'm running the simulation?

Any advice is appreciated, thank you

Answer 1

[Jacob Mathew]

Hey Vincent,

Using Parallel Computing Toolbox should be one of the most effective method to try first as you would be able to parallelize the training. Without actually seeing the code or model, it would be hard to comment on specific approaches but refer to the following documentation as a starting point:

https://www.mathworks.com/help/releases/R2024a/reinforcement-learning/ug/train-agents-using-parallel-computing-and-gpu.html