1 / 10

Hardware Component Discussion group Moderators: Stan Ahalt CASC

General discussion: research computing needs, condominium strategies, cloud computing impact, research group clusters, is network connectivity sufficient ?, need for control of local resources. Paper discussion: Investing in HPC: A predictor of research competitiveness in US academia" by Ahalt

oshin
Download Presentation

Hardware Component Discussion group Moderators: Stan Ahalt CASC

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


    1. Hardware Component Discussion group Moderators: Stan Ahalt (CASC/RENCI) and Julio Facelli (U. Utah) Participants: Dick Sacher (U. Delaware), Bryan Shader (U. Wyoming), Keith Weber (Idaho State), Jonathan Caldwell (West Virginia), Franklin Fondjo (U. Oklahoma), Dana Brunson (Oklahoma State), GaryBozyhnisky (URI), Joel Tohline (LSU/LONI)m Rick McMullen (U. Kansas), Bruce Seger (U. Maine), Dan Andersen (Kansas State)….. (Check spellings and add e-mail)

    2. General discussion: research computing needs, condominium strategies, cloud computing impact, research group clusters, is network connectivity sufficient ?, need for control of local resources….   Paper discussion: “Investing in HPC: A predictor of research competitiveness in US academia” by Ahalt and Apron.   This paper provides quantitative validation of anecdotic evidence that has been widely accepted in the HPC community on the impact of HPC on academic research.

    3. The most relevant findings of the paper relevant to this workshop are:   There is a sweet spot for HPC academic systems at ~20-30 TF   The correlation and causality between HPC investments an research competitiveness

    4. General conclusions:   State/Regional hardware resources are very important. There are technical and political (local control, PR, show off, etc.) arguments to continue provide for these resources.   Investment in HPC resources will stimulate research competitiveness.

    5.   Are HPC hardware investments relevant to the EPSCoR mission?   Yes. If EPSCoR mission is to make States more competitive, the Ahalt-Apron report provides quantitative evidence that hardware HPC investments are a viable strategy to increase States research competiveness.

    6. Are the current EPSCoR programs an adequate to vehicle for these investments   No. The panel did find that none of the current programs, Track I, II and IIc, is adequate to provide EPSCoR investments in HPC hardware.

    7. What are the key components of a potential EPSCoR program to make investments in HPC hardware

    8. The proposed HPC infrastructure should be for general support in computational sciences and its impact on scientific research should be justified by the existing and potential user base and the quality of the science that will be supported by the system. The supported science and activities proposed should be consistent with the State science plan.     The HPC resources should be open to any qualified research program in the jurisdiction(s) regardless of its inclusion of the original proposal. The jurisdiction(s) should establish a transparent allocation resource process for this purpose.       The impact of the EPSCoR investment should be measured in accordance with the general improvement of the research competiveness of the awardees.

    9. EPSCoR should establish a common homogeneous assessment plan for all the awards in this program.   Proposals should be funded for 3 years; they could be renewed but the grantees will be required to provide a long term sustainability plan that will extend the operation of the facility beyond the NSF funding period. The sustainability plan should include strategies for providing resources for operations and periodic hardware refreshments.     Power and environmental direct cost should be allowed.

    10. Personnel for system operations, user support and outreach should be allowed in the proposed budgets, but direct support for research should not be provided under this program.      Matching funds for hardware purchases should not be mandatory.   Multistate collaborations should be encourage, but not mandated.    The panel recommends 6 year program, with an annual funding of $9,000,000 and with a maximum award of $1,000,000 per jurisdiction over 3 years.

More Related