In this video, Susan Coghlan from the Argonne Leadership Computing Facility provides a sneak peek at its upcoming Intel Xeon Phi coprocessor-based supercomputers. As the third of three Coral supercomputer procurements, the deal will comprise an 8.5 Petaflop Theta system based on Knights Landing in 2016 and a much larger 180 Petaflop Aurora supercomputer in 2018. Intel will be the prime contractor on the deal, with sub-contractor Cray building the actual supercomputers.
“Aurora’s revolutionary architecture features Intel’s HPC scalable system framework and 2nd generation Intel Omni-Path Fabric. The system will have a combined total of over 8 Petabytes of on package high bandwidth memory and persistent memory, connected and communicating via a high-performance system fabric to achieve landmark throughput. The nodes will be linked to a dedicated burst buffer and a high-performance parallel storage solution. A second system, named Theta, will be delivered in 2016. Theta will be based on Intel’s second-generation Xeon Phi processor and will serve as an early production system for the ALCF.”
Aurora will deliver more than eighteen times the computational performance of Mira, its predecessor at the ALCF, using a nearly equal number of compute nodes. Aurora will be a many-core system—but with nearly an order of magnitude more processors.
When Aurora is launched, it will be a state-of-the-art system for many-core processing. Its unprecedented capacity and revolutionary memory system will provide capabilities far beyond accelerated compute speed. Highly portable programming techniques already in use on present-day massively parallel systems, including Mira, apply directly to Aurora.
- MPI + OpenMP 4.x programming, which offers portability to commodity Linux clusters as well as CPU-GPU hybrid architectures.
- Intel compilers and optimization tools
- Cray compilers and libraries
- Third-party software including common analysis and visualization tools used in DOE supercomputing centers.