Bristol and Cambridge are to turn out to be residence to 2 of the UK’s latest and strongest supercomputers – Isambard-AI and Daybreak.
Earlier this 12 months, the federal government introduced an funding of £225m to create the UK’s strongest synthetic intelligence (AI) supercomputer with the College of Bristol and Hewlett Packard Enterprise (HPE). The funding is a part of a £300m package deal to create a nationwide Synthetic Intelligence Analysis Useful resource (AIRR) for the nation, which was introduced on the authorities’s AI Security Summit at Bletchley Park this week.
This supercomputer, referred to as Isambard-AI, will join with one other new supercomputer cluster in Cambridge, referred to as Daybreak, which is being co-designed by Intel, Dell Applied sciences and the College of Cambridge.
Daybreak shall be hosted on the Cambridge Open ZettaScale Lab. In keeping with Dell, it is going to be essentially the most highly effective AI supercomputing cloud and can run scientific OpenStack cloud software program developed with UK SME StackHPC.
Dell’s system is predicated on Dell PowerEdge XE9640 servers, 4thGen Intel Xeon Scalable processors and Intel Knowledge Middle GPU Max Collection accelerators.. Dell mentioned, because of liquid cooling and versatile configuration, the server system is well-equipped to deal with the calls for of AI and HPC workloads. In keeping with Dell, direct liquid cooling expertise supplies extra environment friendly and cost-effective cooling than conventional air-cooled methods. On the software-side, the HPC makes use of Scientific OpenStack from UK SME StackHPC, which supplies a completely AI- and simulation-optimised cloud supercomputing software program surroundings. That is mixed with the oneAPI open software program ecosystem and optimised frameworks that assist builders pace up AI and HPC workloads and improve code portability throughout a number of {hardware} architectures.
Paul Calleja, director of analysis computing companies on the College of Cambridge, mentioned: “Daybreak Section 1 represents an enormous step ahead in AI and simulation functionality for the UK, deployed and able to use now. The system performs an essential position inside a bigger context, the place this co-design exercise goals to ship a Section 2 supercomputer in 2024 which can boast 10 instances the extent of efficiency. If taken ahead, Daybreak Section 2 would considerably increase the UK’s AI functionality and proceed this profitable trade partnership.”
The system from HP shall be configured with 5,448 Nvidia GH200 Grace Hopper Superchips, which mix Nvidia’s ARM-based Grace CPU with a Hopper-based GPU optimised for energy effectivity and scale. The {hardware} makes use of HPE Slingshot 11 interconnect and practically 25PB (petabytes) of storage utilizing the Cray Clusterstor E1000 optimised for AI workflows.
Just like the Dell {hardware}, Isambard-AI may also characteristic direct liquid-cooling capabilities as a part of the HPE Cray EX supercomputer design to enhance power effectivity and general carbon footprint influence. HPE mentioned the system shall be hosted in a self-cooled, self-contained information centre, utilizing the HPE Efficiency Optimised Knowledge Middle (POD), and shall be located on the Nationwide Composites Centre (NCC), primarily based on the Bristol and Tub Science Park.
HPE can be collaborating with the College of Bristol on a warmth reuse mannequin, extracting waste warmth from the Isambard-AI system to make use of as renewable power to warmth native buildings, supporting the Web Zero carbon effectivity targets for 2030/2040 as mandated by the UK Authorities.
“Collaborations just like the one between the College of Cambridge, Dell Applied sciences and Intel, alongside robust inward funding, are very important if we would like compute to unlock the high-growth AI potential of the UK It’s paramount that the federal government invests in the appropriate applied sciences and infrastructure to make sure the UK leads in AI and exascale-class simulation functionality. It’s additionally essential to embrace the complete spectrum of the expertise ecosystem, together with GPU range, to make sure clients can sort out the rising calls for of generative AI, industrial simulation modelling and ground-breaking scientific analysis,” mentioned Tariq Hussain, head of UK. Public Sector, Dell Applied sciences.