Fresh off the back of the news that it is to purchase computer chip company Arm, also known as the “crown jewel” of UK tech, for $40 billion, Nvidia has announced that it is building a £40 million ($51.7 million) supercomputer dedicated to healthcare research – right where Arm’s headquarters are located, in Cambridge.
Packing 400 petaflops of AI performance and eight petaflops of Linpack performance, the new Cambridge-1 supercomputer is pitched by Nvidia as the UK’s most powerful supercomputer to date. It is expected to come online by the end of the year, and would rank 29th on the latest list of most powerful supercomputers around the world.
SEE: Managing AI and ML in the enterprise 2020: Tech leaders increase project development and implementation (TechRepublic Premium)
The device will be powered by 80 of Nvidia’s DGX A100 systems, which are designed to build and run AI projects at a large scale – in this case, machine-learning applications designed specifically to advance healthcare research.
Unveiling Cambridge-1 at Nvidia’s GPU Technology conference earlier this week, the company’s CEO Jensen Huang said: “Tackling the world’s most pressing challenges in healthcare requires massively powerful computing resources to harness the capabilities of AI.
“The Cambridge-1 supercomputer will serve as a hub of innovation for the UK, and further the groundbreaking work being done by the nation’s researchers in critical healthcare and drug discovery.”
Researchers will be able to tap a suite of tools developed by Nvidia called Clara Discovery, which features pre-trained AI models and frameworks that are optimized for DGX systems. Leveraging data ranging from imaging to radiology through genomics, Clara Discovery will assist researchers in developing applications such as drug discovery.
Pharmaceutical companies AstraZeneca and GSK are set to start using the supercomputer when it comes online, as well as researchers from various institutions including the London-based Medical Imaging and AI Centre, which works hand-in-hand with NHS Trusts to analyze vast amounts of clinical data and deliver AI-powered solutions to patients.
The organization expects that the extra computational power enabled by Cambridge-1 will reduce diagnosis time for patients, and allow for better identification of appropriate treatments and hospital interventions.
Sebastien Ourselin, head of the school of biomedical engineering and imaging sciences at King’s College London, who leads the AI Centre, explained that the unprecedented volume of both memory and power provided by the new supercomputer will enable researchers to solve data-based problems that were previously impossible to harness.
“We have a huge collection of patient data coming from radiologists, oncologists, blood tests, imaging data and other biomarkers,” he told ZDNet. “To make sense of all that data, you will need huge amounts of memory and large computational power. That’s what the supercomputer is going to open – an opportunity to run those applications that we would never have been able to run on standard computers.”
SEE: Where did the Moon come from? Scientists are using a supercomputer to find out
The AI Centre has just launched partnerships with 11 NHS Trusts, which Ourselin estimates will give researchers access to data related to up to 23 million patients across the country. Using supercomputer-powered AI to analyze this vast array of data could provide much deeper insights into the patient journey, from the moment they consult their GP all the way to their final follow-up appointment, to make better decisions about future diseases.
From finding new drugs to diagnosing serious diseases years in advance: according to Ourselin, Cambridge-1 has strong potential to cause breakthroughs in medical research. “If you have access to huge amounts of computational power, then you can start focusing on asking the right question, rather than focus on the technical limitations,” he said.
Nvidia is finalizing the $40 billion (£31 billion) purchase of Arm from Softbank in a move that is worrying some experts in the field. Arm’s co-founder Hermann Hauser wrote an open letter to the UK Prime Minister Boris Johnson expressing his concern, among others, about the impact that the acquisition might have on jobs in the country, as well as on the UK’s influence in the tech sector worldwide.
Nvidia has said it will invest in an AI research center in Cambridge, which will include an Arm-based supercomputer.
Cambridge-1 will eventually be deployed alongside Arm’s supercomputer in the Nvidia-developed AI center, which the US company has described as a future “hub of collaboration” for AI scientists and start-ups across the country.
Earlier this year, Nvidia unveiled a $70 million (£54.3 million) partnership with the University of Florida to upgrade the institution’s existing supercomputer with the company’s DGX A100 systems. The new version of the device boasts 700 petaflops of AI performance, and is the fastest supercomputer in academia to date.