XSEDE Proposal Writing Webinar

Summary of the XSEDE Webinar "Writing a Successful XSEDE Allocation Proposal" I attended last week

  • The full recorded session can be found here: ​https://meeting.austin.utexas.edu/p3pmvkq0mjg/ .
  • Questions I asked and the speaker's answer:
    1. Research collaborations (Typically how many SUs applied/how many SUs awarded) — Research Collaborations are those large projects with multi-PIs. Sites standard. Typically 15~16 million. Currently the total of all research request is about 800 per year, 4.0 billion SUs per year and 1.8 billion awarded.
  1. is it better to submit a big proposal asking a lot SUs or several smaller proposals each of which asking small amount of SUs? One PI is not allowed to apply with different projects as PI. Recommend to combine different projects from the same group to be one. — sounds like big proposal is OK?
  1. Is there a way we can run scaling-testing for our own code on these new machines? Transfer SUs. Some of the machines are very similar, So you don't have to do scaling testings on all of them. For example SuperMIC ( newest supercomputer funded by NSF, located in LSU, will be in production in April 1st 2014) is similar to Stampede.
  • Important points I catch and we might miss before
    1. Justification of SUs: clear simple calculation, log/simple wall time?
    2. local compute resources in details: referees may know some of your big machines.
    3. research team in details: how may faculties, staffs, postdocs, graduate, undergraduate students. Ability to complete the plan.
    4. Publications acknowledging XSEDE and/or feature stories on XSEDE website: productive, are PI, Co-PIs publish together?
    5. There are groups that are awarded 90% of their request.
    6. Ranch (TACC) and XWFS (The XSEDE-Wide File System) can be requested for storage resources without need to request computing at the same time.

Comments

No comments.