Minutes from 12/10/2013

December 10th, 2013 at 12pm in Room 628 Boyd

Agenda for 12/10/2013

Members Present: Jonathan Amster, Jonathan Arnold, Peter Kner, Margie Lee, Jim Leebens-Mack, David Saltz, and Richard Watson

Members Absent: Martha Carr, and Yiping Zhao

Ex-Officio Members Present: Jessie Kissinger, Thiab Taha, Bob Scott, and Guy Cormier

• The minutes of the December 3rd meeting were unanimously approved.
• Guy presented an overview of the proposed purchase based on a quote from Penguin Computing.
  • The purchase includes the components recommended by the advisory committee in the previous meeting.See attached slides.
  • GACRC users can/will have the option of purchasing nodes for which they will have priority through a "Chinese Menu" of options. So much for a general purpose compute node, so much for a high memory node, etc.
  • Guy explained that it is not yet clear whether the purchase can be single-sourced or whether there will need to be a Request for Proposals. Because it is such a large purchase, it will have to go through multiple layers of bureaucracy. It is Guy's opinion that the quote from Penguin Computing is competitive. The quote is for ~$1.45M.
• The financing of this purchase is as follows: $350K has been allocated to the GACRC for 4 years through FY16 as part of a retention package for Jessie Kissinger. The remainder of these funds, $1.35M, will be used for this purchase, and Tim Chester's office (OVPIT) will provide the remaining ~$100K - $150K. The GACRC will have $250K/year in operating funds (split 50/50 between OVPR and OVPIT) available
  • When the 4x $350K was allocated, it was explicitly mentioned that this money could be used for matching funds for external grants. If this purchase goes ahead as is, that money will not be available.
  • There is an NSF MRI proposal being prepared for the 1/23/2014 deadline led by Jonathan Arnold (with 4 other faculty in chemistry, biology, and physics) for the purchase of a mixture of CPU and GPU nodes. The proposal is for $2M and requires mandatory 30% matching by the university. The money that would be used for the matching is the same as above.
  • There was a discussion of reducing the amount of this purchase so that money would be available for the matching. The comment was also made by Jonathan and others that the current purchase strengthens the MRI proposal. It was also pointed out that the current purchase fills a clear need for the computing community on campus. The zcluster is being used at full capacity. David Saltz said that perhaps funds could be approved for this purchase and "set aside" until the Summer. Then, if the MRI proposal was funded, that part of the purchase could be used as matching.
  • When Tim Chester first put forward the idea of a $1.5M purchase for the GACRC, he asked how would we spend the money if it were available.
  • The Advisory Committee unanimously voted to approve the purchase as is.
• Finally, Jim Leebens-Mack gave a summary of the effort to re-write the Mission, Vision, and Strategic Goals of the GACRC. He made the point that we could simply write up a document to be approved by the advisory committee or we could interview stakeholders and create a document that would reflect the goals and needs of the UGA community and would provide guidance for long-term planning.
  • This led to a discussion of the what the university wide goals for computing are and whether the GACRC is the vehicle for these goals or a component of them, in which case an effort beyond the scope of the GACRC advisory committee should be responsible for defining the university wide goals. At the moment, there is not such body, and Bob Scott recommended that the advisory committee determine the needs of the GACRC user community.
  • The GACRC has a "split personality" in that it is part a high performance computing center and part a more general computing resource. The IT Master Plan refers to research computing as more than just high-performance computing. A well developed mission, vision, and strategic goals document would presumably clarify the role of the GACRC.

File:GACRC AC Dec10.pdf