SAN DONATO MILANESE - Eni makes a new technological leap in supercomputing with the launch of HPC7, the system that enters among the most powerful supercomputers on the planet and strengthens the company's European leadership. Thanks to the combination with HPC6, computing capacity exceeds the Exaflop threshold for the first time, opening up new prospects for energy, artificial intelligence and decarbonisation. A milestone that places the Italian group among the world leaders in digital innovation applied to industry.
6th place overall in the new global TOP500 ranking (link)
Eni positions itself as the world's leading company for computing power in the new global TOP500 ranking. Thanks to the combination of HPC6 and HPC7, Eni's computational capacity exceeds the Exaflop threshold. Eni announces the launch of the new HPC7 (High Performance Computing - HPC) supercomputing system which, with a capacity of more than 861 Pflops/s, ranks 6th overall in the new global TOP500 ranking (link), the second supercomputer in Europe, and confirms itself as the most powerful High Performance Computer in the world for industrial use. HPC7 therefore surpasses HPC6, launched in November 2024, which itself entered the TOP 10 by placing 8th in the TOP500 ranking. The combination of the HPC6 and HPC7 computing systems exceeds the Exascale level: HPC6 and HPC7 together are capable of performing more than 1 Exaflop/s (1 Exaflop/s = 1000 PFlops/s), that is, more than 1 billion billion complex mathematical operations per second.
A fundamental step in Eni's strategy for enhancing energy sources and decarbonisation
The achievement of the Exascale class by Eni's supercomputing system represents reaching the most advanced and extraordinary technological frontier in the world of supercomputers and confirms the company's leadership in the sector. The launch of HPC7 represents a fundamental step in Eni's strategy for enhancing energy sources and decarbonisation, a model in which technology is a central element of innovation, capable of driving growth, efficiency and competitiveness in both traditional and transition businesses. Advanced computing and HPC systems continue to be central for Eni, enabling the integration and enhancement of expertise and applications throughout the entire value chain: from subsurface understanding to the optimisation of industrial plants, through to improving the accuracy of geological and fluid dynamics studies for CO2 storage and the development of advanced energy technologies.
A crucial enabler for the internal development of artificial intelligence use cases dedicated to Eni's businesses
Supercomputing also helps accelerate the evolution of the main innovation pathways, supporting the efficiency of emerging supply chains – such as biofuels – and the simulation of complex phenomena, including plasma behaviour in magnetic confinement fusion. Within this framework, HPC is also establishing itself as a crucial enabler for the internal development of artificial intelligence use cases dedicated to Eni's businesses. This technological ecosystem also represents a distinctive value for Eni as it is capable of attracting new initiatives and talent from outside the company, as already demonstrated with HPC6 through the Call4Innovators. In detail, Eni's new HPC system adds 571 PFlops/s to the 477 sustained PFlops/s of HPC6, corresponding to peak values of 606 PFlops/s for the former and 861 for the latter. HPC7 is based on an architecture designed using the same technology that powers the most powerful systems currently operating in Europe and worldwide, combining CPUs and GPUs in a hybrid configuration, with more than 3,400 computing nodes and almost 14,000 GPUs, in order to maximise computational performance and energy efficiency.
The transition towards energy from traditional and renewable sources that is increasingly secure, accessible and clean cannot take place without profound technological evolution
The combined computing power of HPC6 and HPC7 therefore reaches 1048 sustained PFlops/s and 1467 peak PFlops/s. With a value of 65.426 GFlops/W, HPC7 also achieved an excellent ranking in the specific Green500 classification, which measures machine efficiency, placing eleventh worldwide and first among machines in its category. Eni Chief Executive Officer Claudio Descalzi said: «The transition towards energy from traditional and renewable sources that is increasingly secure, accessible and clean cannot take place without profound technological evolution. The adoption of supercomputing and predictive technologies across every activity is fundamental to the development of new energy solutions, reducing emissions, maximising efficiency in exploration and production, and generating value. In this context, the construction and commissioning of HPC7 in an extremely short timeframe, even shorter than HPC6, which was already a benchmark, represents a concrete example of our execution capability: the result of the expertise, commitment and quality of our operational teams.
The Green Data Centre confirms its characteristics thanks to an innovative liquid cooling system
This solid digital ecosystem, developed through talent, collaboration and internal research, not only accelerates our journey towards Net Zero, but also consolidates our strategic positioning and competitive advantage in the market». Both supercomputers are located in a dedicated area of Eni's Green Data Centre, benefiting from infrastructure designed to combine operational efficiency and environmental sustainability. The Green Data Centre, already among Europe's leaders for energy efficiency and emissions reduction, confirms its characteristics thanks to an innovative liquid cooling system.
Some explanations for better understanding
1. HPC7 uses HPE Cray EX4000 technology and HPE Cray ClusterStor E2000 on HPE's AMD architecture.
2. Computing Power: the system has achieved computing power of more than 861 PFlops peak (Rpeak) and more than 571 PFlops sustained (Rmax), placing it among the most advanced infrastructures in the world.
3. Size: the system comprises 3,480 computing nodes, incorporating a total of 13,920 GPUs.
4. Node Composition: each node in the system consists of four AMD Instinct™ MI300A Accelerated Processing Units (APUs), each combining 24 AMD EPYC™ Zen 4 CPU cores and 228 CDNA3 GPU compute units, for a total of 96 CPU cores and 912 compute units. Each node has 512GB of HBM3 memory.
5. High-Performance Network: the HPE Slingshot network ensures fast and reliable interconnection between nodes, facilitating high-speed data transfer (200 Gbps).
6. Cooling System: the system employs Direct Liquid Cooling (DLC) technology, enabling the dissipation of 96% of the heat generated.
7. Electrical Power Consumption: the system has a maximum electrical power consumption of 9.4 MW (including cooling and support systems), giving the machine an efficiency rating of 65.426 GFlops/W.
Articoli correlati: