BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Prototype of a Parallel Analysis System for CMS using PROOF
DTSTART;VALUE=DATE-TIME:20060214T152000Z
DTEND;VALUE=DATE-TIME:20060214T154000Z
DTSTAMP;VALUE=DATE-TIME:20130619T060324Z
UID:indico-contribution-267@cern.ch
DESCRIPTION:Speakers: Dr. GONZALEZ CABALLERO\, Isidro (Instituto de Fisica
  de Cantabria (CSIC-UC))\nA typical HEP analysis in the LHC experiments in
 volves the processing of data\ncorresponding to several million events\, t
 erabytes of information\, to be analysed in\nthe last phases. Currently\, 
 processing one million events in a single modern\nworkstation takes severa
 l hours\, thus slowing the analysis cycle. The desirable\ncomputing model 
 for a physicist would be closer to a High Performance Computing\none where
  a large number of CPUs are required for short periods (of the order of\ns
 everal minutes). Where CPU farms are available\, parallel computing is an 
 obvious\nsolution to this problem. Here we present the tests along this li
 ne using a tool for\nparallel physics analysis in CMS based on the PROOF l
 ibraries. Special attention has\nbeen paid in the development of this tool
  to modularity and easiness of usage to\nenable the possibility of sharing
  algorithms and simplifying software extensibility\nwhile hiding the detai
 ls of the parallelisation.\n\nThe first tests performed using a medium siz
 e (90 nodes) cluster of dual processor\nmachines on a typical CMS analysis
  dataset (corresponding to root files for one\nmillion top qurk pairs prod
 ucing fully leptonic final state events distributed\nuniformly among the c
 omputers) show quite promising results on scalability.\n\nhttp://indico.ce
 rn.ch/contributionDisplay.py?contribId=267&sessionId=9&confId=048
LOCATION:Tata Institute of Fundamental Research D406
URL:http://indico.cern.ch/contributionDisplay.py?contribId=267&sessionId=9
 &confId=048
END:VEVENT
END:VCALENDAR