Project

General

Profile

FrameworkWorkshopIdeas2011

Logistics

Co-organize with CERN? Yes.

Who?
Infrastructure development experts:

  • SLAC - Geant4 people, anyone like Norman Graf or Tony Johnson?
  • CERN - let John Harvey decide
  • ATLAS - Paolo
  • CMS - Chris Jones

Experiment people:

  • CMS - Liz
  • ATLAS - Paolo
  • Geant4 - ?
  • Intensity Frontier experiments - ?

When?
Late November? Thanksgiving week.

Purpose and Goal

  • Understand where we think the world is at with regards to frameworks
  • What direction are groups going? (target architectures, techniques, and features)
  • Who is doing what for prototyping and for new experiments?
  • How do we synchronize work activities? (is it necessary at all?)

Looks like we should have 1/2 the workshop be talking with the experiments about needs and requirements, and the other 1/2 to be discussions amongst framework building experts (including experiment developer team members). We should take the position that this is forward looking work (beyond the current code base to some extent). Buy-in could mean experiment input, but with a new model that does not necessarily mean running existing code. This new model will be ready for the new architectures and systems as they are needed. This may not fit in with Pere's plan of attack.

Know that we are looking for overlaps in our program of work and the CERN team's projects. This program includes the studies of restructuring the G4 toolkit/framework, and also satisfying the needs of the smaller intensity frontier experiments, particularly in the area of DAQ systems with little to no hardware trigger.

Notes

Interesting aspects that should be included in the goals

  • Requirements / constraints on G4 and the other event frameworks that are to be considered for re-engineering (on a timeline for readiness)
  • Target software architecture for G4 and other event frameworks that are to be considered for re-engineering (on timeline for readiness)
  • Target hardware for these things

Should we do a WBS session for the G4 and art framework projects? Yes, in the way we did the JDEM and mu2e DAQ ones. This means use the chart method, where horizontal axis is release/time, vertical axis is subsystem, cell are features that will be available for this release / feature.

Problem with Pere's proposed agenda - No invite to SLAC (for Muon Collider or other thing).

Segments (no order yet)

Current state-of-the-art

  • What frameworks are available now and what are the variants?
  • What are the advanced features available?
  • What makes them good?
  • Where are they weak?

Current and future users

  • survey of future experiments

Requirements

  • future computing systems
  • complexity of data
  • distributed processing needs
  • event organization
  • processing speed
  • parallelism
  • configuration
  • online use
  • programming language(s) supported for user code
  • How general a framework is appropriate?
    • "event processing" only?
    • visualization components?
    • interactivity?
  • ...

Future framework

  • toolkits, techniques, and libraries that are available