Choose timezone
Your profile timezone:
EP R&D Software Working Group Meeting
→
Europe/Zurich
32/1-A24 (CERN)
EP R&D Software Working Group Meeting
Zoom Meeting ID
62893115044
Host
Graeme A Stewart
Alternative host
Andre Sailer
Useful links
Join via phone
Zoom URL
Software R&D Working Meeting Minutes
2022-05-18
Apologies:
Room: Graeme, Andre, Anna, Dalila, Erica, Paul, Placido, Valentin, Pere
Remote: Alberto, Javier, Andi, Marco, Juraj
Introduction
- For the June R&D days, we have asked that the software slot is on Monday
- Andi found funds to buy a machine with an Nvidia GPU for tracking
- This is a “CAD Workstation”, ordered and deliverec in 1 day
- For GPUs, we need to cover development, CI and performance testing
Turnkey Software
- What is a Frame? Is it orthogonal to an event?
- Events have a fixed set of collections
- Frame can gather collections, for some sbset of things that are in the event store
- Challenge of deciding where analysis code should live - as RAnalysers or as part of Key4hep
- An example is a reimplented vertex finder, implemented in RDataFrame instead of Gaudi
- Need to provide documentation and worked examples
- For the validation, are you sure the input events are exactly the same? (Slide 17)
- Yes, this could be an issue, should ensure they are exactly the same
- e.g., generate only LCIO, and from there convert to EDM4hep
- Data model converter should only be a difference of rounding!
- Start validating single events, one by one!
- Yes, this could be an issue, should ensure they are exactly the same
- Any idea of the overheads in the conversion?
- About 2%, but it depends a lot on the number of conversions
- For schema evolution, there should be a discussion between the RNTuple developers and PODIO
There are minutes attached to this event.
Show them.