RAISE members presenting at the CERN openlab annual technical workshop.
On March 16-17 2023, CERN openlab held its annual technical workshop at the CERN site in Geneva, Switzerland. CERN openlab is a unique public-private partnership between CERN and leading tech companies, which works to drive innovation in the computing technologies needed by CERN’s research community. The workshop was organized and chaired by Maria Girone, CoE RAISE WP4 leader. At the event, Maria Girone was announced as the new head of CERN openlab. Girone, who has served as CERN openlab’s Chief Technology Officer since 2016, recently received a prestigious Italian award and founded the Swiss Chapter of the Women in High-Performance Computing advocacy group.
The ambitious upgrade programme for the Large Hadron Collider (LHC) poses significant computing challenges. When the High-Luminosity LHC, (HL-LHC) comes online in 2029, around ten times the computing capacity of today will be required. Simply spending more money to buy more equipment is not an adequate solution, instead, IT experts across CERN are working to find ways to do things smarter. About 30 R&D projects are carried out through this collaboration, addressing challenges related to the next generation of supercomputers, known as “exascale”, as well as artificial intelligence (AI), and quantum computing. CERN openlab also runs projects aimed at sharing knowledge and expertise with research communities beyond particle physics. All these projects were presented at the two-day technical workshop, which was held in the CERN Council Chamber.
Figure 1: CERN openlab workshop at the CERN Council Chamber (Image: CERN)
There were two talks given by members from CoE RAISE on the work done at CERN in Task 4.1 of WP4, which is on Event Reconstruction and Classification at the CERN HL-LHC. The first, given by David Southwick, covered heterogeneous High-Performance Computing (HPC) benchmarking for Exascale, including both compute-based benchmarking as well as data-driven benchmarking regarding aspects such as data transfer and I/O. The second talk, given by Eric Wulff highlighted an area where AI and HPC work particularly well together, distributed Hyperparameter Optimization (HPO). Wulff’s talk covered not only the use of large-scale supercomputer centers for HPO but also the use of a special kind of Quantum Computer, a so-called quantum annealer (QA), working together with classical computing systems in a hybrid quantum-classical HPC workflow. This way of working is enabled by the modularity of the infrastructure available at the Jülich Supercomputer Centre (JSC) which hosts both classical HPC resources and quantum machines, allowing users to access both technologies through the same programming environment.
The event was attended by 145 people (in person and online), including representatives of member companies Intel, Oracle, Siemens, Micron, Google, IBM, Roche and Comtrade. As well as discussing ongoing projects, the workshop provided an excellent opportunity for considering emerging challenges and identifying opportunities for mutually beneficial collaboration.
CERN openlab article