Skip to content

Default --tasks-per-job settings #264

@aebid

Description

@aebid

Currently, in the framework each tasks has nJobs on scale:

Production
InputFileTask (nDataset) -> AnaTupleFileTask (~nTotalFiles) -> AnaTupleFileListTask (nDataset) -> AnaTupleMergeTask (~nTotalFiles) -> [ optional AnalysisCacheTask (~nTotalFiles) ]

Plotting
HistTupleProducerTask (nTotalFiles) -> HistFromNtupleProducerTask (nDataset * nVariables) -> HistMergerTask (nVariables) -> HistPlotTask (nVariables)

This means that the number of jobs per task varies greatly. To reduce batch jobs, generally you can run with tasks-per-job 10 for tasks until HistMergerTask. After that, HistMergerTask and HistPlotTask can be run with tasks-per-job 1.

Suggestion is to add a default value for tasks-per-job, making it easier to simply submit HistPlotTask while still being reasonable to nJobs for batch submission.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions