COMPASS-ABS: Reducing Fragmentation in Shared GPU Clusters for Deep Learning Training Workloads
SIF is introduced, a metric built on the notion of partial-nodes that is independent of historical workload knowledge that matches both theory and production and COMPASS-ABS is proposed, which employs the COMPact-ASSured (COMPASS) algorithm to confine the cluster state within a tight Anchor-Based Space (ABS).
Yu-Kai Zhou, Hong-Fan Wu
· 0 citations