Among the major issues that the European Commission has had and must address, in addition to those of work, climate, immigration, Brexit management and in the last year that of the health emergency due to the coronavirus epidemic, there is is the so-called strategic autonomy also linked to technological innovation: that is, the ability of the Union, and its member states, to maintain control and production, and to establish standards and rules, of all the technologies that are necessary for growth and development, but also for the safety of its citizens. The European Union has historically been dependent on mostly North American and Asian technology companies, but it has long been trying to find areas in which it can achieve global hegemony, or at least try to compete with other powers. For example, it has launched some initiatives to autonomously develop an ecosystem for high-performance computing.
In 2018, the European Commission promoted, within Horizon (its main research funding program), the European Processor Initiative (EPI) project, the aim of which is to design and implement a plan for the production of a new family. of European processors with low power consumption for high performance computing (in English High Performance Computing, HPC), necessary for the so-called supercomputers (i.e. with high computing performance).
The project was carried out by a consortium of 28 partners, including companies, research centers, universities from ten different European countries, carefully selected to ensure the necessary key design skills, and to ensure that these shared skills remain in Europe. Among these is E4 Computer Engineering, founded in 2002 in Scandiano, in the province of Reggio Emilia. E4 deals with the design, construction, verification and installation of systems specifically intended for use in HPC applications, such as numerical modeling (the process of numerically simulating a physical process), the analysis of large amounts of data, artificial intelligence and machine learning techniques.
High-performance computing takes place through the use of “supercomputers”, which in the IT field are defined as “clusters” (cluster) of computational nodes, ie computer systems, servers and other resources closely connected and able to function as a single system, performing parallel processing. Supercomputers are used to process large amounts of data in many fields: scientific, government, industrial. In the future, experts expect further growth in the volume of data to be analyzed. The EPI project therefore arises from the contextual need to limit energy consumption while maintaining high computational performance for future systems called exascale.
The processing power of a system is in fact measured in FLOPS (floating-point operations per second): to give an order of magnitude, currently the fastest systems work in the order of petaflops (1015 flops), that is, millions of billions of operations per second. The EPI project aims to create processors that support the so-called “exascale” machines, i.e. computers and computer systems capable of processing billions of billions of operations per second (exaflops, i.e. 1018 flops) and maximizing performance at the lowest level of consumption. energy.
The goal is therefore a low-power processor optimized for exascale HPC, in order to increase computation processing speed and efficiency, and which can be used by programs currently in use in industry and research. As part of its participation in EPI, E4 is responsible for designing, building and testing the accelerator testing module (a component of the processor), which forms an integral part of the processor, and the processor power components.
EPI aims to contribute to the achievement by 2023 of European independence and competitiveness in the field of HPC, providing a processor capable of satisfying the requirements of various sectors that require high computing intensity, from the development of Intelligence. Artificial to Big Data processing up to the implementation of Autonomous Driving systems in the automotive sector.
