Research Group Quotas

Namespace quotas

Namespace	GPU slices	CPU	Memory	Notes
rg-compsci	2	8	32 Gi	Computer Science
rg-abi	1	4	16 Gi	Auckland Bioengineering Institute
rg-general	shared	shared	shared	No guaranteed allocation
pais-inference	2	—	—	Reserved for vLLM inference workloads

GPU slices refer to time-sliced L40S allocations (1 slice ≈ 12 GB effective VRAM).

Checking current usage

kubectl describe resourcequota -n rg-compsci
kubectl top pods -n rg-compsci

Requesting a quota increase

Contact the PAIS team with: - Your namespace - The resources you need (GPU slices, CPU, memory) - Expected duration and workload description

Groups that contribute GPU hardware to the platform receive priority quota.