jsun-m

any updates on this?

> The following paper from 2022 describes in more detail the "device proxy" approach proposed in Balancing Efficiency and Fairness in Heterogeneous GPU Clusters for Deep Learning mentioned above: [Singularity:...

I don't think this will work out of the box? Currently we force all GPU's on the agent to be filled up to the max with 1 GPU workers.

@aaydin-tr Thanks for you suggestion! That makes sense. It does seem like many releases have small individual changes, which makes it hard to see the bigger picture. I’ll discuss this...