Cerebras Systems AI supercomputer selected to rapidly accelerate AI research on EIDF

Cerebras Systems, the pioneer in high performance artificial intelligence (AI) compute, and EPCC, the supercomputing centre at the University of Edinburgh, have announced the selection of the world’s fastest AI computer, the Cerebras CS-1, for EPCC’s new international data facility for the Edinburgh & SE Scotland City Region. Featuring the HPE Superdome Flex Server from Hewlett Packard Enterprise, this leading-edge AI deployment enables the next wave of natural language processing (NLP) and data science research for public, private and academic sectors across the City Region and the UK.

“We are proud to announce this audacious infrastructure investment and partnership with the world leaders in AI computing. This installation will enable massive breakthroughs in our vision for data science and greatly accelerate our research across genomics and public health, including time-sensitive and pressing issues such as leveraging AI across large models to advance COVID-19 therapeutic research.”

Professor Mark Parsons

EPCC will employ a unique combination of the Cerebras CS-1 system, powered by the high-performance Wafer Scale Engine (WSE) processor, and an extremely large-memory HPE Superdome Flex Server system for unprecedented AI scalability and massive data handling capability.  This advanced AI supercomputing system will greatly reduce training time, the most time-intensive part of AI, enabling many more ideas to be tested, and will be available in the new Edinburgh International Data Facility (EIDF) for academic researchers and data scientists in the public and private sectors.

The EIDF will support the University of Edinburgh's ambition to deliver more data-driven innovation industry partnerships and transform the region into the 'Data capital of Europe,’ one of its key ambitions as a partner in the City Region Deal.

EPCC’s mission is to accelerate the effective exploitation of novel computing throughout industry, academia and commerce. The CS-1 deployment amplifies and extends the reach of that mission for AI applications with local- to global-scale impact. Specifically, this partnership will accelerate AI-powered data science initiatives in the Edinburgh and Southeast Scotland City Region, enabling national-scale genomics research for public health initiatives. It also advances AI research for natural language processing of global academic and industry interest, using deep learning models like BERT in the School of Informatics’ Institute for Language, Cognition and Computation.

“We are excited to bring our industry-leading CS-1 AI supercomputer, coupled with HPE’s advanced memory server, to EPCC and the European market to help solve some of today’s most urgent problems. Our vision with the CS-1 was to reduce the cost of curiosity, and we look forward to the myriad experiments and world-changing solutions that will emerge from EPCC’s regional data centre.”

Andrew Feldman

The CS-1 is built around the world’s largest processor, the WSE, which is 56 times larger, has 54 times more cores, 450 times more on-chip memory, 5,788 times more memory bandwidth and 20,833 times more fabric bandwidth than the leading graphics processing unit (GPU) competitor. In AI compute, large chips process information more quickly, producing answers in less time. Depending on workload, from AI to HPC, the CS-1 delivers hundreds or thousands of times more performance than legacy alternatives, and it does so at a fraction of the power draw and space. The CS-1 has won numerous awards including Fast Company’s Best World Changing Ideas, IEEE Spectrum’s Emerging Technology Awards, Forbes AI 50 2020, HPC’s Readers’ and Editors’ Choice Awards and CBInsights AI 100 2020.

EPCC will use the HPE Superdome Flex Server, a powerful and easy-to-use high performance front-end storage and pre-processing solution for the CS-1 AI supercomputer. This will enable users to employ large datasets and application-specific pre- and post-processing of data for AI model training and inference on the CS-1, allowing the CS-1s WSE to operate at full bandwidth. HPE Superdome Flex Server will be robustly provisioned with 18 terabytes of memory, 102 terabytes of high-performance flash storage, 24 powerful Intel Xeon CPUs, and 12 network interface cards to deliver 1.2 terabits per second of data bandwidth to the Cerebras CS-1.

“HPE has a long-standing collaboration with EPCC to develop solutions to some of the most challenging computational problems, and we are excited to be working at this time to provide a highly productive AI platform. By tightly coupling a Cerebras Wafer Scale Engine with a HPE Superdome Flex Server In-Memory host, we are aiming to enable researchers to tackle complex AI workloads at unprecedented rates.”

Mike Woodacre

Cerebras Systems https://cerebras.net