February 18, 2010

Performance Meeting - 2010/02/18
Phone Number: (812) 856-3600
PIN:          002690#

Agenda:
Subversion repository moved here.
  • Active Milestones:
  • Benchmarking - Framework - Setup a provenance server to storing benchmark results
  • Mike Lowe (IU) created VM for us - gw41.quarry.iu.teragrid.org or inca.futuregrid.org
  • Performance: Instrumentation - Select timing library (Netlogger, OTF)
    • Not started
  • Performance: Benchmarking - Bare OS - HPCC - Create configuration files
  • UCSD iDataplex has power and should have networking on Fri -- Jenett scheduled to setup on 22/23
  • Performance: Benchmarking - Framework - Write Inca reporter to execute benchmarks
    • Not started
  • Inca: Benchmarking - Framework - Acquire machine and setup Inca server
  • Mike Lowe (IU) created VM for us - gw41.quarry.iu.teragrid.org or inca.futuregrid.org

################
Performance Meeting - 2009/11/12
Number:  812-856-3600
Participant code:  002690#

Agenda:  
  • SC'09
  • Review of architecture writeups
  • benchmarks (Robert)
  • benchmark framework (Shava)
  • benchmark repository (Shava)
  • instrumentation (Raphael, Thomas)
  • performance portal (Shava)
  • user tools (Robert)
 
Notes:
 
Attendees:  Shava (UCSD), Robert (IU), Warren (TACC)
 
  • SC'09
    • Decided to do partner specific posters - UCSD and TACC have customized posters
    • Meeting next week at 9am on Thursday at Ambridge Event Center
      • will send details on room name
      • phone confirmed (ACTION: Warren: can you send me a number to reach you) 
      • Thomas will give short presentation on Vampir
    • Robert to attend SPEC meeting -- will find out about licensing restrictions and whether we can run the benchmarks on FutureGrid -- may need to go to SPEC legal
  • Performance documents have been moved over to "Working Documents" -- old links still work too (must be logged in to see content)
  • Shava added "Deliverables" with time estimates and "Dependencies" to frameworksection and to software monitoringsection (ACTION: all: to do similarly for their sections so we can define our first cycle - 3 month timeframe likely)
  • Warren discussed distributed messaging system for experiment harness -- could be used to get environment information during an experiment and archive it
    • tie-in to Pegasus which will be used to actuate experiment -- details to be worked out
    • tie-in to Vampir which will also incorporate environment information into traces (ACTION: Warren to look at Vampir) 
    • tie-in to Inca which could send notifications to experiments via the messaging system
  • New members from IU (ACTION: can you please send me an email and let me know your interest and level of participation)