Jan 2 – 7, 2018
Skeikampen, Norway
Europe/Oslo timezone

Advances in software and computing for HEP

Jan 5, 2018, 3:30 PM
Skeikampen, Norway

Skeikampen, Norway

Hotellvegen 3, 2652 Svingvoll, Norway
Invited Detectors, computing and software Friday PM


Graeme Stewart (CERN)


High energy particle physics faces huge challenge in software and
computing in the coming decade. At the LHC ALICE and LHCb will make
major upgrades for Run 3, requiring much improved software, while the
challenges faced by ATLAS and CMS for High Luminosity LHC imply that
software and computing must improve by around x10 beyond technology
evolution alone to fit into the anticipated resources. Belle II and
DUNE will also start their data taking with comparable rates. This
problem is made all the harder to solve because developments in
computing hardware do not provide the simple Moore's Law scaling of
the past. Concurrency, multi-core and vector, has become essential for
achieving performance on CPUs. Adapting many pieces of HEP code that
were developed decades ago for this is a major challenge. GPUs and
FPGAs are now increasingly used and require a complete rethinking of
many algorithms and software frameworks. In each case software
sustainability also needs to improve.

To tackle this challenge the HEP Software Foundation launched a
Community White Paper process to lay out a roadmap for HEP computing
evolution into the 2020s. I will outline the process that was
undertaken to gather input from across the HEP community. I will look
at the main areas of work, from event generation to analysis, and
survey the research and development projects for the coming years. Finally, I will conclude
on how the community can turn the white paper proposals into a genuine
"software upgrade" process that matches the ambition of the rest of
the HEP experimental program.

Primary author

Presentation materials