Transition to cluster 7

Background

As you may know, we currently manage several clusters that are functionally independent. These are

• cluster 2 - this is what we often refer to as "gimel" and includes both CPU and GPU managed partly by slurm and partly by SGE
• "the Keiser cluster" - this is the machines bequeathed by Mike Keiser, and currently heavily used by his remaining students.
• cluster 0 - this is the pre-Toronto cluster that includes features such as DUDE, supported by ZINC12.

During the fall of 2024, we are moving to consolidate many of these machines into a new cluster "Cluster 7" which we hope will become the new default cluster. We hope it will be more maintainable, and better in every way.

The first step of a long journey starts with a single step. One of the first steps will be moving "mk-gpu-3" from Keiser cluster to Cluster 7. This will happen in the days soon after Monday Oct 21.

Once cluster 7 is ready for use, you will be invited to use it. Our aim is to make the transition as smooth and painless as possible. We will be working with you individually to help you maintain an optimal level of compute performance.

If you have any concerns, please write your sysadmin team on slack or otherwise.

-- John Irwin, Oct 21, 2024.