Mellanox Accelerates Supercomputers and More

Mellanox announces that the Oak Ridge National Laboratory Summit supercomputer is on target for completion early this year. The system uses dual-rail Mellanox EDR 100Gb/s InfiniBand to interconnect thousands of compute nodes containing IBM POWER CPUs and NVIDIA GPUs, delivering 200Gb/s network speed to each of the compute platforms.

CORAL (Collaboration of Oak Ridge, Argonne and Lawrence Livermore National Laboratory) is a project initiated in 2013 to develop the technology that would address the Department of Energy’s future computing needs. After an evaluation, the collaboration between Mellanox, IBM and NVIDIA was selected by the CORAL project team for the Summit supercomputer in Oak Ridge National Laboratory and Sierra supercomputer at the Lawrence Livermore National Laboratory.

“The advanced InfiniBand capabilities will significantly enhance the ability of applications to effectively communicate latency-sensitive data,” says Jim Rogers, computing and facilities director for the National Center for Computational Sciences at ORNL.

In related news, Mellanox InfiniBand plans to accelerate Japan’s fast supercomputer for artificial intelligence applications. Mellanox Technologies reports that Fujitsu has selected InfiniBand solutions for the National Institute of Advanced Industrial Science and Technology (AIST)’s the AI Bridging Cloud Infrastructure (ABCI) supercomputer, a fast supercomputer in Japan, and large-scale open artificial intelligence infrastructure.

The ABCI will be deployed by Fujitsu at the University of Tokyo and is scheduled to begin operations in the second quarter of 2018. The dual EDR 100G InfiniBand network will provide an aggregated data throughput of 200Gb/s, connecting the Fujitsu PRIMERGY servers, including multiple NVIDIA GPUs and NVMe storage. The Fujitsu system is expected to deliver a theoretical peak performance of 550 petaflops in half-precision floating point operations, or 37 petaflops of double-precision floating point operation performance.

In further related news, a collaboration with Lenovo will power Canada’s large supercomputer center for HPC applications. Mellanox Technologies announces that the University of Toronto has selected the Mellanox InfiniBand solutions to accelerate its new leading supercomputer in Canada. The new system will leverage the new Dragonfly+ network topology, built on Lenovo’s ultra-dense ThinkSystem SD530 high-performance computing (HPC) server for processing power, developed to deliver an Exascale-ready infrastructure optimized for cost performance and scalability.

In additional related news, InfiniBand connects two of the top five supercomputers. InfiniBand solutions, according to Mellanox, accelerate 77% of the new HPC systems on the TOP500 list deployed since the previous list (June 2017 to November 2017). InfiniBand accelerates 60% of the total HPC systems on the list and two of the top five supercomputers, including reportedly the fastest supercomputer in the world and the fastest supercomputer in Japan.

The TOP500 List has evolved to include hyperscale, cloud, Web 2.0 and enterprise platforms, in addition to the HPC and machine learning systems.

For more info, visit Oak Ridge National Laboratory, Lawrence Livermore National Laboratory and Mellanox.

Sources: Press materials received from the company.