Staff Priorities - Scott: Difference between revisions

From SCECpedia
Jump to navigationJump to search
Scottcal (talk | contribs)
No edit summary
Scottcal (talk | contribs)
No edit summary
 
(452 intermediate revisions by the same user not shown)
Line 1: Line 1:
== Yesterday ==
== Current Items ==
*Queued USC with random - GPU code failing, must investigate
*Modify CyberShake to work on RSQSim ruptures
*Investigated slow PreSGT - slow reading of files
**Modify PreSGT step to determine list of points to save SGTs for from RSQSim rupture files
*Sent Efecan sample coscheduling scripts for testing on Titan
*CyberShake training
*Discussed Palmdale hotspot with Kevin
*Update CyberShake study wiki pages to point to SQLite files
*Create wiki page for CyberShake RSQSim, and upload sample seismograms


== Today ==
== Project Backlog ==
*Check on USC random GPU failures
*Run CyberShake post-processing on RSQSim events, once DB is populated
*Speed up PreSGT for 200 m surfaces
*Produce candidate site list for Bay Area CyberShake
*Work on Summer School science talk
*Investigate improved load-balancing for DirectSynth SGT handlers
*Get timing data with rupture variation generator to Yifeng
*Experiment with OMP parallelization for post-processing
*Run test to see if locality matters
 
== Blocked ==
*No
 
== Follow-ups ==
*No
 
 
 
*Try out SeedMe so I can respond to Amit
 
 
 
== Areas of Responsibilities ==
* CyberShake
* Broadband CyberShake
* Code migration

Latest revision as of 17:56, 1 February 2018

Current Items

  • Modify CyberShake to work on RSQSim ruptures
    • Modify PreSGT step to determine list of points to save SGTs for from RSQSim rupture files
  • CyberShake training
  • Update CyberShake study wiki pages to point to SQLite files
  • Create wiki page for CyberShake RSQSim, and upload sample seismograms

Project Backlog

  • Run CyberShake post-processing on RSQSim events, once DB is populated
  • Produce candidate site list for Bay Area CyberShake
  • Investigate improved load-balancing for DirectSynth SGT handlers
  • Experiment with OMP parallelization for post-processing