AMD and Cray to build world's first exascale supercomputer

Dubbed Frontier, the AMD/Cray supercomputer will offer 1.5 exaflops of processing power for Oak Ridge National Laboratory

AMD and Cray are to build what they claim will be the world's first exascale supercomputer for the US Oak Ridge National Laboratory (ORNL). The machine will offer more than 1.5 exaflops of processing power and be delivered in 2021. The contract for the supercomputer is worth more than $600 million, according to Cray.

The system will be based on Cray's latest Shasta architecture and Slingshot interconnect, while AMD will be supplying its Epyc server microprocessors, running alongside AMD Radeon Instinct GPUs. A custom high-bandwidth, low-latency Infinity Fabric will connect four AMD Radeon Instinct GPUs to one AMD Epyc CPU per node.

In addition, AMD and Cray are developing an enhanced version of the open source ROCm programming environment to run on the machine, and enhancing a number of other programming tools to support the new machine.

Cray CEO Peter Ungaro said that the system "will be composed of more than 100 Cray Shasta cabinets with high density compute blades".

Ungaro continued: "The new accelerator-centric compute blades will support a 4:1 GPU to CPU ratio with high speed AMD Infinity Fabric links and coherent memory between them within the node. Each node will have one Cray Slingshot interconnect network port for every GPU with streamlined communication between the GPUs and network to enable optimal performance for high-performance computing and AI workloads at exascale."

The purchase is part of the CORAL-2 procurement by the US Department of Energy (DOE), encompassing a $1.8 billion Trump administration effort to build the world's first exascale supercomputer.

ORNL is the DOE's largest multi-programme science and energy laboratory. According to AMD, the researchers at ORNL will use the Frontier system's computing power and support for artificial intelligence techniques to simulate, model and improve understanding of the interactions underlying the science of weather, sub-atomic structures, genomics, physics, and other scientific fields.

As things currently stand, IBM enjoys the top two spots in the Top500 list of the world's most powerful supercomputers, last updated in November 2018. The 2.4 million core IBM Summit supercomputer, operated by ORNL, claims top spot. It was installed in June 2018. It is followed by the 1.6 million core IBM Sierra in second place.

The Sunway TaihuLight and Tianhe-2A take third and fourth spot. Over the next two years, supercomputer makers in both China and the US will be competing to be the first to design, make and install the world's first exascale supercomputer.

Japan's Fujitsu, meanwhile, is also developing an ARM-based supercomputer to use in its forthcoming Post-K exascale effort, which is being developed by Fujitsu for the Riken research institute in Wakō, Japan.

Delta is a new market intelligence service from Computing to help CIOs and other IT decision makers make smarter purchasing decisions - decisions informed by the knowledge and experience of other CIOs and IT decision makers.

Delta is free from vendor sponsorship or influence of any kind, and is guided by a steering committee of well-known CIOs, such as Charles Ewen, Christina Scott, Steve Capper and Laura Meyer.

Ten crucial technology areas are already covered at launch, with more data appearing and more areas being covered every week. Sign-up here for your free trial of the Computing Delta website.