Memory Upgrade

We have added more memory to 29 nodes on the Snowstorm cluster. It is distributed as follows:

• 16GB on 8 nodes, compute-10-9 to compute-10-2
• 8GB on 21 nodes, compute-10-1 to compute-8-1
• 4GB on the rest as before

We hope to see a significant increase in throughput as we have seen that some jobs have had
increased memory requirements lately. No extra action is needed. Just make sure that your
memory specifications is as accurate as possible, and the scheduler will take the appropriate action
to make sure that you get the space that you ask for.