With both data center electricity and hardware prices skyrocketing, most organizations will soon be looking to squeeze more efficiencies from their current investments, especially those embarking on resource-heavy AI projects running on Kubernetes.
Over the past year, a cross-foundational working group from the Cloud Native Computing Foundation had been fiendishly developing an enhancement to the Kubernetes scheduler that would allow users to be far more specific in how they allocate jobs to CPUs, network cards, GPUs and various AI…
