Minutes for the ITC-Research Computing Management Meeting
March 26, 2007 at 10:30AM
2015 Ivy Road, 2nd Floor Conference Room

Members: Alice, Hamp, Jim, Joe S., Mark, Mike, Robin, Terry, Tim S., Tom, Tim T.

Attending: Alice, Hamp, Jim, Joe S., Mark, Mike, Robin, Tim S., Tom, Tim T.

Chair: Tim T. Recorder: Alice Howard

I. Connection Question: What about research computing do you find compelling and why? (or not)

II. Corrections to minutes from last meeting on February 26, 2007? None.

(Click to review the agenda from this meeting )

III. Old Business:     (Italics indicate original agenda item text)

• Proposed research computing hardware expenditures. Decided at February meeting:
• Buy large memory nodes (~8 nodes of ~16GB) for new cluster to replace Athena. Have operational in mid-summer.
• Could use any remaining funds to buy additional nodes for dogwood
• Retire birch, not buy new nodes to make "new birch" cluster. Likely not offer infiniband (or other high speed interconnect) replacement.
• Hamp recommended that we go with Xeon (dual core) and distributed some sample pricing from Dell for three different configurations. After discussions about power needs, 16 GB vs. 32 GB, etc., the group chose the:

Xeon 5160 (80 w)
3.0 GHz, 4 MB cache
32 GB (8 * 4 GB DIMM) - maximum memory available
160 GB disk (mirrored)
$15,418.25
head + 4 compute = $ 77,091.25
head + 8 compute = $138,764.25

(plus $7500 for other hardware – i.e. Nortel switch, PDU, storage)

• Applications that are likely to be used on this cluster include commercial packages like Ansys, Matlab, Mathematica.
• Researchers can buy-in to cluster with 1-2 nodes and get dedicated access.
• However for Dogwood, the policy is still: buy 5, get dedicated access to 4. The original thinking was to “contribute to the common good” – but it might be time to consider changing to ‘buy 1, get 1” or “buy 5, get 5”, etc.
• Review and modify queuing policies that discourage rapid turnaround over long running jobs – want to experiment with policies -- and have future public forum on queuing policies. Increase awareness on existing test queues of clusters.
• from prior meetings: (Tom) Genomics might want to buy in to cluster, but probably not want 5 nodes at our current specs – Tom will talk to them. They need large memory computers.
• What about allowing researchers to buy "priority" time on dogwood cluster? By “priority” do we mean a pre-emptive jump to the head of the line/queue? Or the ability to “own/reserve” a queue? We might be able to find a solution from someplace else – Tim T. will look into it – at least we could use it as a straw-man.

• Linux Clusters: Birch, Cedar, Dogwood - any issues, concerns or upcoming updates/changes?
• Matt Neurock’s group might still want to keep Birch running – will take one of their jobs – run it on Birch and Dogwood -- and see where it runs faster.
• Aim to shut Birch down by August 1 – will need to see how Neurock’s code performs on Dogwood – meanwhile other Birch users can be moving to Dogwood starting now.
• Any update on conversion to authenticate to Eservices via Kerberos for Unix logins/connections for Blue.unix? (AKA Phase 2 of Netbadge CDP and "unified password" web site). No change.
• Infrastructure Supporting Research Task Force (https://www1.seas.virginia.edu/itrtf/). (Maillist is itrtf@virginia.edu)
• UVa's participation in ORNL's response to NSF RFP for next generation/iteration of Supercomputing Center . The full proposal was due on 2/2/07 - Mike to check with Mitch on status/update?

IV. New Business:

• Project Implicit, PI: Brian Nosek, Psychology Department. Project description at: NIH-IPP.researchplan.B-Nosek.pdf
• Looking for kind and amount of assistance from ITC (mostly) for project. Project web site is at Harvard. Uses Oracle back-end. Gets tens of thousands of hits, especially when featured on TV. Have Java programmers who would welcome consultations on how to make code more efficient and robust. Would be interested in moving infrastructure (Oracle DB and cluster) here. Web site is http://implicit.harvard.edu
• Carla Lee, Andrew Sallans (from Brown SEL) and Tim Tolson met with Brian and two students on the project last week to get some background on the project and their needs. Brief summary of this meeting
• Do we want to host this? What does it need? What are they asking from us? (Note: we do not have any spare power or space) Add Hamp into the next meeting with them.
• In light of our familiarity with collaborative model of communities, consider for next meeting, re-examining our purpose, our membership and our constituency.
• what are we about/doing?
• who are we as group?
• Do we have the right group of folks?
• Are we organized/operating effectively/efficiently?
• We could re-examine these issues – is there an opportunity for a community here? A way to get faculty input?
• Any other Items?

V. Adjourn by 12:00 and next meeting

• Next Scheduled Meeting of ITC ResComp Standing Committee: Monday, April 29th, 10:30 AM in ITC-2015 Ivy Road, Room #102 (First Floor Conference Room)!

Please send suggestions, additions, corrections to: Tim Tolson or Alice Howard

Go to: ITC Research Computing Committee Home Page