-
Notifications
You must be signed in to change notification settings - Fork 108
JobSplitting Argorithms
ticoann edited this page Apr 12, 2017
·
7 revisions
The goal of job splitting is to make the job length for optimal use of resources (~8 hours) - There are various parameters are used to calculate approximate job length (most importantly TimePerEvent)
Basic splitting is defined by events_per_job is this is set in the Spec/Splitting Algorithm, if this value is not set it will be calculated by TimePerEvent (depending on split alogrithm EventBased, EventAwareLumiBased)
when evnets_per_job is none calculate using timePerEvent
- First determine how many lumis can be in the job depending on various conditions. max_events_per_lumi (default 20K) if inputfile has only one lumi and avgEvtsPerLumi (events in the file/lumis in the file) is bigger than max_events_per_lumi - fail this job (on creation). events_per_job