By Zoe Cournia, Rossen Apostolov and Richard Norman
(Based on presentations and discussions held at the EuroHPC Summit 2023 in Gothenburg on 21-23 March 2023)
What does it mean to make good use of exascale HPC, what are the challenges for the European CoEs and what are the differences for each application area? These are some of the questions posed during the panel on user communities’ expectations led by Rossen Apostolov, BioExcel CoE director.
Life Sciences are at the dawn of exascale HPC, which will allow scientists to model whole cells (or viruses) in atomistic detail, simulate entire metabolic pathways or systems using quantum chemistry, leverage the power of AI to its full potential, reach higher accuracy and spatiotemporal resolution with biological systems and allow for a much more predictive approach to drug creation which will reduce the time and cost of bringing drugs to the market. The potential benefit to society of exascale computing in the Life Sciences is huge but many challenges remain, especially around data generation and management, training and support and access to infrastructure, especially for industry.
Exascale HPC applications are data intensive and generate terabytes of data every month. How should we store, manage and analyse these data, and how should we make them accessible in the longer run? For this we need new architectures, networks, software and workflows for archiving, analysing and collecting the data.
Knowledgeable High Level Support Teams in the field of Life Sciences are urgently needed to help the support, development and optimization of codes in the field. Software engineers working in Life Science applications need incentives, varied career paths and recognition with permanent jobs.
Leadership and skills in HPC use will be crucial to developing a digitally-skilled workforce in the use of complex infrastructure such as HPC. There is a need for multiple stakeholders in the HPC value chain to play their part; from Centres of Excellence and National Competence Centre to initiatives such as EOSC, as well as industry and academia. Full digitisation of training services and the appointment of HPC Ambassadors will be key to facilitate access and ensure a broad outreach, especially in European regions which are less well represented in the HPC landscape. Training life scientists, computer scientists and chip designers together will ensure better integration and faster progress in developing the appropriate HPC skills.
These efforts coupled with investments and co-design in software will be crucial for maintaining Europe’s leadership in HPC applications. Strengthening the cooperation between industry and academia in addition to stronger industrial investment in HPC will be a huge driver for the European economy. More incentives and training for industry on the use of HPC will be key in achieving this. Industrial users will have the need for different access models including more urgent and interactive (short time access) computing and more secure access to ensure confidentiality and IP protection. Additional initiatives such as ‘Fortissimo’ and ‘FF4Europe’, which help SMEs access HPC infrastructure and benefit from HPC simulations, should be established as well as broader private-public partnerships and consortia.
The future of European exascale HPC looks bright, but we are still some years away from demonstrating its impact to society as a whole.
(no AI-based research tools were used in the generation of this article)