Italian energy company Eni is upgrading its HPC4 system with new gear from HPE that will be installed in Eni’s Green Data Center in Ferrera Erbognone (a province in Pavia, Italy), and delivered “as-a-service” via HPE’s GreenLake platform. The new HPC4, spanning more than 1,500 heterogenous HPE ProLiant DL385 Gen10 Plus nodes, offers ~20 percent more computing power with a peak performance of about 22 petaflops. 

The system’s 1,522 ProLiant nodes are equipped with two AMD second-generation Epyc “Rome” 24-core 7402 CPUs and a mix of GPU devices, all connected by Nvidia HDR InfiniBand. Comprising the bulk of the system, there are 1,375 nodes that house two Nvidia “V100 O&G” GPUs; 125 nodes have two Nvidia A100s; and 22 nodes include two AMD Instinct MI100 GPUs. Additionally, there are 20 CPU-only login nodes.

The new contract with HPE provides a complete update to HPC4, the HPE ProLiant DL380 Gen10 system that was installed in 2018. Benefits reported by the partners include a roughly 20 percent improved computing capacity for running simulations, and a doubling of storage capacity to refine the accuracy of image-intensive modeling and simulations of complex energy research.

The new infrastructure leverages the Cray ClusterStor E1000 storage system and the HPE Data Management Framework “to support complex, image-intensive workloads in modeling and simulation.”

Eni told HPCwire that the combination of Eni Green Data Center and HPE GreenLake will boost sustainability by mitigating over-provisioning of resources and preventing the wasted energy consumption that goes along with that.

HPE said the new infrastructure will improve energy usage and reduce electronic waste by using HPE Asset Upcycling Services. “This is part of the circular economy initiative from HPE Financial Services, which leverages asset longevity to reuse products, by recycling equipment from its existing HPC4 system and replacing it with newer solutions,” the company reported.

The original HPC4 installation (let’s call it HPC4-OG) was the largest announced energy supercomputer when it debuted in 2018. It entered the June 2018 Top500 list at number 13, delivering 12.2 Linpack petaflops out of 18.6 peak petaflops. HPC4-OG spanned 1,600 ProLiant DL380 nodes, each equipped with two Intel 24-core Skylake processors and two Nvidia Tesla P100 GPUs. The system ranks 33 on the current Top500 list.

What about HPC5?

Eni still operates the HPC5 supercomputer, built by Dell and launched in February 2020. Housed in Eni’s Green Data Center, HPC5 spans 1,820 Dell EMC PowerEdge C4140 servers, each with two Intel Gold 6252 24-core processors and four Nvidia V100 GPUs. Server nodes are connected by Nvidia 200Gbps HDR InfiniBand. The system entered the Top500 list in June 2020 with 35.4 Linpack petaflops out of a theoretical 51.7 petaflops, becoming the world’s fastest industrial supercomputer. Currently in ninth position on the list, HPC5 still holds that distinction.

In April of last year, early on in the pandemic when it was ravaging Italy, Eni announced that its HPC5 system was contributing to COVID-19 research, as part of part of the European EXSCALATE4CoV project. Eni worked with Cineca to provide the consortium with its technical skills and its HPC5 computing power, and in November 2020 enabled one of the most complex molecular supercomputing experiments ever conducted.

Eni’s Green Data Center houses all of the company’s supercomputing systems and data. Eni states that the Green Data Center was developed to achieve world-leading energy efficiency, partly due to the nearby photovoltaic plant that provides the facility with a portion of its power.

HPE reports that its GreenLake business has more than $5.2 billion in total contract value and about 1,200 enterprise customers in 50 countries. HPE recently won a GreenLake HPC-as-a-Service deal with the U.S. National Security Agency (NSA) valued at $2 billion.