Sustainable Infrastructures

Heterogeneous computing platforms and infrastructures

CERN openlab contributes to this area of innovation via collaboration with resource providers for access to test resources on cloud and HPC infrastructures. The main focus is on understanding how to access and integrate external resources into the computing workflows employed by the CERN community (e.g., IT, experiments). This can be done via direct access to cloud environments or pilot collaborations with cloud providers and HPC sites. This activity includes optimisation of AI workflows on large-scale HPC systems.

Impact: The expected impact is the co-development of adapted access models to HPC and cloud infrastructures, integrated and optimised workflows, including AI, and costing models.

Computer architectures and software engineering

CERN openlab contributes to this area of innovation via collaboration with resource providers for the provision of hardware and software components, dedicated specialised expertise from the technology partners, funds for engineers and developers, as well as training and education opportunities. The main focus is on assessment, benchmarking and validation of accelerated architectures (e.g. GPUs, FPGAs) and new processors (e.g. RISC V) following user requirements. 

Impact: The expected impact is to provide the HEP community with innovative technology and key expertise on ongoing and future research programmes.

Storage and data management

CERN openlab contributes to this area of innovation via collaborations with technology providers and systems integrators. This includes, for example, the evaluation of new storage media, the co-development of specific functionality for multi-disciplinary applications, or the definition and implementation of data workflows for HEP in the context of the emerging data analysis facilities concept. 

Impact: The expected impact is to provide the HEP community with storage and data management tools that can effectively support cutting-edge research programmes.

Artificial intelligence algorithms, platforms and applications

CERN openlab contributes to this area of innovation by fostering collaboration with technology providers and research institutions developing state-of-the-art platforms, services, and methodologies. This includes access to software and expertise, as well as large-scale testbeds for co-creating new AI models and optimization workflows. The areas of work include distributed AI optimization, generative AI, foundation models for physics, as well as optimal deployment of AI-based algorithms on modern computing architectures, and benchmarking of these architectures on AI workloads. Work is also ongoing in the realm of real-time AI inference (on edge or accelerator devices) as part of the detector data acquisition, fast data selection, and accelerator control.

Impact: The expected impact is to enable the science community to access and leverage AI resources and skills.

Applications for society and environment

Energy-efficient computing and AI for society are of great interest to technology providers, data centres, and software developers. CERN openlab will contribute both in the establishment of dedicated investigations in green computing, life and environmental science, and in raising awareness through explicit requirements in the definition of development projects.

Impact: CERN openlab will support the migration to energy-efficient architectures and develop tools for sustainable data centres. Furthermore, it will keep its work with other sciences and contribute to research with a societal impact.