Named after the English mathematician Ada Lovelace, the mainframe boasts 58 nodes and 2,230 cores – the equivalent of 58 computers each with 40 processors – 800TB of storage and is housed in a secure room on campus.
It replaces Hydra, which was introduced in 2010.
Lovelace, designed by tech company OCF, consumes roughly half the electricity of its predecessor whilst delivering three to four times the output.
The new machine was installed last year. However, the unveiling has been postponed due to the coronavirus pandemic.
It is currently being used by more than 200 academics, PhD students and researchers across the University for a variety of projects, including motorsport aerodynamics, quantum physics, next-generation battery modelling and fluid dynamics.
Matthew Cook, assistant director of infrastructure and operations, said: “As one of the country’s top 10 research-led universities, there are approximately 100 academic papers per year connected to the use of the University HPC facilities.
“We therefore need a HPC service that gives us the ability to change scientific methodology quickly rather than waiting for it to compute. Which was becoming a problem with Hydra.
“Lovelace is a huge step forward for us. The computational capacity knows no bounds.
“The users are primarily from our science and engineering departments, but as we are actively encouraging a University-wide HPC culture and the cluster becomes more integral to our diverse teams, we have increased uptake from other areas such as creative arts and economics.
“We are making academic colleagues aware of what services are available and providing examples of the art of the possible.”
Credit: º¬Ðß²ÝÊÓƵ
The new HPC cluster provides 58 Lenovo SD530 compute nodes with 40 compute cores each, 2,320 total, three dedicated GPGPU nodes each with dual NVIDIA Tesla V100 GPU’s and two high memory visualisation nodes.
All systems are connected via Mellanox HDR100 InfiniBand fast networking and backed by 800TB of Lenovo DSS-G Storage which uses the IBM Spectrum Scale FileSystem.
Software in this deployment includes Bright Cluster Manager - Advanced Edition, together with Ansible for Cluster Management, Moab HPC Suite for Workload Management and the Intel Parallel Studio XE Cluster Edition suite of development tools.
Cook added: “This is not just about upgrading an HPC.
“This is about designing an environment where we can support diverse applications and create research that matters.
“We know that to attract the best researchers to our Excellence 100 programme, which in turn increases our chances of more research grants, we need to listen to user needs.
“We therefore built in new GPGPUs graphic cards into the system to increase optimisation to improve research outputs.”
Credit: º¬Ðß²ÝÊÓƵ
Dr Pooja Goddard, senior lecturer in Chemistry at º¬Ðß²ÝÊÓƵ, runs collaborative research projects with other universities, as well as industry, on computational materials modelling with several PhD students and postdocs.
She said: “We conduct ground-breaking research including testing for next-generation design of battery, solar and nuclear materials, as well as more far-reaching applications such as modelling fingerprint development reagents and their reactions with surfaces.
“Having a high-spec facility like Lovelace allows us to perform cutting-edge simulations putting us at the forefront of materials modelling research.
“It also helps hugely when applying for grant funding, to say that we have up-to-date in-house high-performance computing facilities.”
“And it is important for recruiting top-class students and developing their skills and confidence to become the next generation of computational scientists.”
Russell Slack, managing director at OCF, added: “We’re really pleased that we have been able to offer the technical expertise and support capabilities to help the University to achieve better research impact to generate new knowledge and insight.
“Lovelace will help the University to continue to attract world-class researchers, as well as grants to continue with their research excellence.”