Functions§
- direct_
gpu_ 🔒processes - Keep only NVML processes whose GPUs resolve to one NUMA node.
- expand_
gpu_ 🔒processes - Expand NVML processes to peer processes in the same exact cgroup.
- expand_
gpu_ 🔒processes_ with - fit_
gpu_ 🔒processes - Fit workload hints into the BPF map without partially expanding a cgroup.