Hey guys, I’m new to Ray and wondering the usecase below works with Ray.
I have two laptopcs (mac) and want to utilize these gpus to create a ray gpu cluster. I want to put one mac parent node and the other one worker node connected to the parent node. Is this feasible?
anyone care to answer my question?
One thing to consider is the Ray Accelerator Support, check the GPUs you are trying to use are supported there. Assuming you took care of the network settings and both laptops can connect to each other, one idea could be to use the on-prem Ray setup to configure your Ray cluster.
Somehow related, In this repo I use a similar approach to configure two Lambdalabs Cloud instances to run a Ray Training Job. The steps after setting up the infra might work as a reference for you.
1 Like