Webinar 2025 What exactly is a Resource Allocation Competition (RAC) Compute Award?

The Resource Allocation Competition (RAC) gives researchers the opportunity to request and be awarded quantities of storage, cloud and/or compute resources. In the case of storage and cloud, the resources awarded can be directly translated into what is made available to the researcher on the cluster (e.g. a 100TB storage allocation translates directly into a 100TB quota on a file system). For CPU or GPU compute awards, however, the translation of CPU or GPU years awarded to what is made available to the researcher is not as direct of a translation. Rather, a RAC compute allocation is implemented as a target share of the compute resources on a cluster for an account. The target share is a key factor in determining the priority of the account’s jobs in the queue based on the recent usage achieved by the account. This presentation describes how compute allocations are implemented on the clusters, covering topics like the relationship between Rapid Access Services (RAS) compute shares and RAC allocations, expected vs unexpected wait times, and monitoring the progress of using a compute allocation.