In general QOS:

• It is not exclusive
• If you do not use all available cores, the other cores are open for others to use
• Max Memory per core: 1.2GB

In large QOS:

• Jobs are exclusive
• When you request a node, it is all yours
• Can use less cores per node in order to get more memory per core
• Can request more memory per core (assuming you don’t go over the total memory of the compute node