Zoom Link:
https://cern.zoom.us/j/67245514307?pwd=QkNIYzNhbXJrYWxsWTZ6ZVFFb1pMQT09
Meeting ID: 672 4551 4307
Passcode: 920424
## Attendees
Rob Ross, Peter Van Gemmeren, Serhan Mete, Patrick Gartung, Torre Wenaus, Rui Wang, Amit Bashyal, Chris Jones, Adam Lyon, Kyle Knoepfel, Shane Snyder, Philippe Canal, Ken Herner, Liz Sexton-Kennedy, Lisa Goodenough, Doug Benjamin, Barnali Chowdhury
## Management News
*No Paolo, no Salman, no news.*
FY23 plan from AHM seems clear, no negative feedback, carry on.
## Darshan
Shane: Working to get a new Darshan release out, likely next week, including some fork() safety fixes. Also look at autoconf for working with desired HDF version (Ken Herner).
Peter: Next step to look at derivation? More I/O intensive.
Rui: No updates this week. Plan for more runs.
Amit: To make Darshan work with ATLAS/Athena, is there anything special we need to do? No we think it should work now. Working on HDF5/ATLAS and want to see Darshan output.
Doug: Kerstin ok'd someone working on Darshan in the analysis space. Need to find a time to sync up with Shane. First thing to do is collect some typical examples, start with Open Data stuff for CMS, "DOD PHYS" for ATLAS?
Peter: More familiar with upstream processes. Derivation could be good.
*Discussion of common scenarios, need to downselect.*
## HDF5 / ROOT serialization
Amit: Not HDF5 exactly, but testing TMPIFile with ROOT serialization. Working to simplify tutorial code so it runs, work around some error.
Philippe: May be that something wasn't built. Missing an auxiliary file that isn't part of the tutorial itself. Jet event is in the "tree" tutorial.
*Discussion of how to proceed related to TMPI code.*
Peter: Integrated writing to HDF5 into ATLAS, however it's not production ready at this time. Amit is applying lessons learned into the code, cleaning it up.
## GPU friendly data model
Amit: No presentation this week. One of the data structures for the 2D case is being rethought. Early in development. Hopefully something to show next meeting.
Peter: Still inspired by DUNE? Yes, ProtoDUNE inspiration.
Kyle: Should we come up with a plan for surveying other efforts? Amit has gotten some hints/guidelines, but we could consider more speakers, more thorough survey.
Peter: Definitely can have speakers present to us on offloading. Peter knows ATLAS people, is confident there are CMS people. Most of those people are in Europe, time zone challenges, would maybe need to have one-off meeting(s) that is/are more amenable.
*Will collect names.*