This page interprets the results of the production tests in terms of the measured running time per event and the projected running time per file and for an ensemble of chains representing a typical production group. Finally we repeat this with the measured sizes of files.
The results section contains information on the tests used to make these projections, their current status and a detailed break down of their results. That page should be read before interpreting any of the information below.
The times shown here and used to make projections are the integrated user and system CPU time for a given tier divided by the number of events run over. This averages any overheads, such as ART job initialisation, first event instantiations, data-base calls and job finalisation, over the course of the tier. In addition tiers are run sequentially as part of the same job, this reduces overheads for later tiers relative to running in isolation. As such these times are to be taken with a pinch of salt. In order to be in any way representative the test jobs must have been configured to run over a number of events typical of a single job in the production running.
The graphs shown here are interactive. You can click on any tier on the legend to toggle it on or off.
The projections were made using the assumptions detailed in the table below. Where a tier is not explicitly mentioned it follows the pattern of the default tier of it's chain. The efficiencies quoted should reflect that measured in the results section of this test.
|Chain||Tier||Number of events to produce||Efficiency [%]||Number of events to run||Number of events per file|
|FD data NuMI||default||6,700,000||100||6,700,000||69|
|FD data cosmics||default||27,500,000.0||100||27,500,000||1,702|
|FD genie FHC nonswap||default||1,500,000||100||1,500,000||3,000|
|FD genie FHC swap||default||1,500,000||100||1,500,000||3,000|
|FD genie FHC tau||default||1,500,000||100||1,500,000||3,000|
|FD genie RHC nonswap||default||1,500,000||100||1,500,000||3,000|
|FD genie RHC swap||default||1,500,000||100||1,500,000||3,000|
|FD genie RHC tau||default||1,500,000||100||1,500,000||3,000|
|ND data NuMI||default||1,900,000||100||1,900,000||2,184|
|ND data cosmics||default||8,000,000||100||8,000,000||8,956|
|ND genie FHC nonswap||default||40,000,000||100||40,000,000||3,000|
|ND genie FHC nonswap||rock||1,280,000||4||32,000,000||800|
|ND genie RHC nonswap||default||40,000,000||100||40,000,000||3,000|
|ND genie RHC nonswap||rock||1,280,000||4||32,000,000||800|
Similarly we project the size of the final output files based on the measured size per output event.