Today Mellanox announced that InfiniBand has been chosen to accelerate the new Eagle supercomputer at NREL. The pending 8 Petaflop system will consist of more than 2100 nodes and deliver 3 times the performance compared to NREL’s current supercomputing platform. Eagle will be put into production in January 2019.
InfiniBand’s smart In-Network Computing acceleration engines, RDMA technology, and high reliability and robustness will ensure the highest performance, efficiency and scalability for NREL’s HPC, AI, storage, cloud and other applications,” said Gilad Shainer, vice president of marketing at Mellanox Technologies. “We are happy to be part of the NREL new supercomputing platform, helping to advance the leading edge of energy research.”
Eagle is configured to run compute-intensive and parallel computing jobs. It is a cluster comprised of 2114 nodes (servers) that run the Linux operating system (Red Had Linux or the derivative CentOS distribution), with a peak performance of 8 PetaFLOPS. The nodes are connected to each other and to storage by a high-speed 100Gb/s EDR InfiniBand network. All nodes and storage are connected using an enhanced 8-dimensional hypercube topology that provides a bisection bandwidth of 26.4 terabytes/s. Eagle has NFS file systems for home directories and application software as well as a 14 petabyte high-speed Lustre file system for parallel I/O.
The Eagle supercomputer will enable NREL’s researchers to run increasingly detailed models that simulate complex processes and phenomena, to gain new insights and innovations in energy efficiency and renewable energy technologies,” said Steve Hammond, NREL’s Computational Science Director at NREL.