Project milestone Open Access
The current technologies used for HPC systems will not be able to sustain the performance increase requested by the HPC/HPDA/AI1 application communities. Since the end of the Dennard scaling and with the approaching end of the Moore’s law, the standard CMOS2 technology has to be complemented by other approaches if we want to continue to deliver more performance.
The three main HPC ecosystems beside Europe, namely US, China and Japan have undertaken significant research initiatives to work on these new approaches. The investigations include topics such as new materials that are more efficient than CMOS, new architecture, photonics, analogue system or quantum technologies. The level of investment is high in these three countries but we do not believe that they have already a competitive edge over Europe.
This white paper presents first why is it mandatory to look for new HPC technologies. After a short presentation of the international landscape, we review the research options that could sustain the development of HPC. This analysis is the result of a work organized by EXDCI-2 to connect the HPC communities with the European electronics and photonics experts.
The last section presents some recommendations for Europe to strengthen its position in technologies relevant for future HPC system. These recommendations could be implemented through EuroHPC actions and/or initiatives of the future Horizon Europe programme.
EXDCI-2-White paper Future HPC technologies v3.pdf