Minutes of the ITC-Research Computing Standing Committee

Meeting on October 28, 2002

Members: Alice, Bill, Brian, Dawn, David, Ed, Hamp, Kathy, Jim, Mark S., Martha, Michael, Robin, Sue Ellen, Steve, Terry, Tim S., Tom S.

Attending: Alice, Bill, Brian, David, Ed, Hamp, Kathy, Jim, Mark S., Martha, Robin, Steve, Terry, Tim S., Tom, Bob Reynolds

Chair: Tim T.

Recorder: Alice

I. Discussions/Old Business:

1. Aspen Cluster rollout: the announcement is ready; the web page is up (http://www.itc.virginia.edu/research/aspen_cluster) – it is modeled on the SP2 page and includes a tutorial for the new cluster. It was decided to go ahead with announcing that the new Aspen cluster is ready to be used

2. Review of quotes from Aspen Systems and options for additional nodes: had some further discussion about whether or not to invest Bill Pearson’s $50K in myrinet nodes. Bill P. would like us to buy some myrinet nodes and find out if anyone will use it. What if we do get an additional $100K in ETF funds (which are currently loaded in our budget) – does the decision about how to spend the $50K then have ripple effects on how we might best spend the additional $100K/ETF funds? Most of our (and SEAS) current use/need is for large serial jobs. What criteria should be used in making the choice: performance? who might use the new nodes? The group concluded that Tim T. should ask Aspen Systems for a quote for 32 nodes (Xeons with myrinet) and a 64 port chassis. And we will figure out if we need to get parallel versions of existing software packages – subtract these software costs out, and spend the rest on nodes, etc.

3. Once the Aspen Cluster is generally available, should we down the IBM SMP box and put PBSPro on it? This might not be as simple as it sounds and there was concern about the login process. It was decided to find a box that could function as a front-end login -- and to move to AIX 5.2.

4. Migrating Engineering school users off kalman, kelly, frank, flemming, landau: It would be a good idea to tie this migration to the demise of the SP2 and coordinate the migrations with the Engineering School (Mitch Rosen). Should get the list of users of the E-school systems – plus the list of SP2 users – check for overlaps – and focus on finding the right home for each user. The goal for moving users off kalman, kelly, etc., is Summer 03. The Engineering School is getting a new 48 node cluster targeted for high memory users.

Next regular meeting: Monday, November 25, 2002 at 10:30 AM in 117 Astronomy