Job Sizing

JEDI generates jobs based on the following task parameters:

Name	Description
nFilesPerJob	The number of input files per job
nGBPerJob	The total size of input/output files and working directory
nEventsPerJob	The number of events per job
cpuTime	HS06sec per event
cpuEfficiency	CPU efficiency (0.9 by default)
baseTime	The part of the job execution time not scaling with CPU power
outDiskCount	The expect output size per event
workDiskCount	The working directory size

If one of the first three parameters n*PerJob is specified, jobs are generated accordingly. If some computing resources cannot accept those jobs due to resource limitation, such as small scratch disks and short walltime, the brokerage avoids those resources.

If they are not specified, jobs are generated to meet the limitation of each computing resource. The number of events nEvents in each job must satisfy the following formulae:

\[S \geq inputDiskCount + max (0.5 GB, outDiskCount \times nEvents) + workDiskCount\]

\[W \geq \frac {cpuTime \times nEvents} {C \times P \times cpuEfficiency} + baseTime\]

where S, W, C, and P are the scratch disk size, the wall time limit, the number of CPU cores, and the HS06 core-power at the computing resource, respectively. inputDiskCount is the total size of job input files, a discrete function of nEvents. Note that inputDiskCount is zero if the computing resource is configured to read input files directly from the local storage resource.