Skip to main content

Cluster Partitions

Partitions and Quotas

Use sinfo -l to view the cluster queue information and node status.

Partition NameResource TypePriorityPriceUser QuotaDefault Job Run Duration
acd_uGPU(ACD)Shared (Low)Low TierCPU: 128 coresGPU: 16 cards7 days
acd_ueGPU(ACD)Exclusive (Medium)Medium TierCPU: 128 coresGPU: 16 cards7 days
emergency_acdGPU(ACD)Emergency (High)High TierCPU: 128 coresGPU: 16 cards7 days

Job Extension

Users submit an IT work order through HKUST GO to apply for an extension of the job running time (only one extension is allowed, and it shall not exceed 7 days).