In order to enhance their role in the field of artificial intelligence, Intel, Dell, and Cambridge University have announced their collaboration in developing the joint project of the supercomputer “Dawn Vaz 1”.
Specialized technical teams worked on creating the fastest giant artificial intelligence computer in Britain, which combines the power of artificial intelligence and high-performance computing to address some urgent global challenges.
The giant computer Dawn will bring Britain closer to reaching the threshold of Quintillion operations per second. This threshold is known as Exaflop One (or as Exascale).
In order for the reader to understand the immense processing power of the Exascale system, everyone in the world would need to perform calculations for 24 hours a day for over four years just to achieve one second of processing power in the Exascale system.
Regarding this announcement, Adam Ro, the Chief Technical Officer for High-Performance Computing at Intel in Europe, the Middle East, and Africa, said: “Dawn significantly enhances the scientific computing and artificial intelligence capabilities existing in the UK, it is currently operational and working at the Open Zettascale lab at the University of Cambridge. Dell’s PowerEdge XE9640 servers provide a suitable platform for hosting the Intel Data Center GPU Max accelerator, enabling a wide range of choices through the oneAPI application programming interface. I am very excited to see the sciences that this giant computer can deliver, and I am enjoying enhancing the Open Zettascale lab partnership between Intel, Dell, and the University of Cambridge, expanding its scope to the scientific and artificial intelligence communities in the UK.”
The Director of Research Computing Services at the University of Cambridge, Dr. Paul Calleja, highlighted the importance of the Dawn Phase 1 project in developing artificial intelligence and simulation capabilities in the UK. This project is now ready for use and represents a significant step towards providing a giant computer in the second phase, with the performance of this giant computer in the second phase expected to be ten times better by 2024. If the second phase is achieved, it is expected to greatly enhance artificial intelligence capabilities in the UK, in addition to continuing this successful industrial partnership.
The ‘Dawn’ giant computer is the fastest in the UK and relies on artificial intelligence, and will support key workloads in the UK in academic research and industry. Its areas of use will include healthcare, engineering, green homogeneous energy, climate modeling, cosmology, and high-energy physics.
The giant computer Dawn will be joined by the Fugaku 1 supercomputer and the Dawn supercomputer to the team. The recently announced formation of the AIRR facility, a national facility in the UK aimed at helping researchers enhance artificial intelligence capabilities. Dawn, supported by UKRI, will significantly increase simulation and computational power for artificial intelligence in the country, both in basic research and industrial applications, accelerating research discovery processes and boosting cognitive economic growth in the UK.
The new giant computer relies on Dell PowerEdge XE9640 servers, and with the diverse configuration options and liquid cooling technology, the server system is well-equipped to handle the high workload requirements of artificial intelligence and high-performance computing. Direct water-cooling technology provides a more efficient and cost-effective solution compared to traditional air cooling systems.
Each PowerEdge XE9640 server in this system combines two processors from Intel Xeon Scalable fourth-generation processors and four Intel Data Center GPU Max accelerators to deliver powerful performance and high efficiency in solving real-world scientific problems.
This is integrated with the system ecosystem through the oneAPI application programming interface to help developers accelerate workloads for artificial intelligence and high-performance computing and enhance their ability to transfer code instructions between multiple hardware architectures.
Participants in the Supercomputing 23 conference in mid-November in Denver, Colorado will be able to access technical details and performance figures of the giant computer Dawn Phase 1.
