Supercomputers and high performance computing (HPC) play an important role in speeding up calculations and the analysis of data, which could previously have taken many years to complete.
The value and importance of HPC at universities has grown significantly over the past 10 years. It’s no exception at the University of Bristol, which has invested more than £16 million in HPC and research data storage in the past decade.
In May 2017, the University hosted a launch event for its new supercomputer, named ‘BlueCrystal4’ or ‘BC4’ for short.
Meeting the needs of researchers and scientists and the projects they work on, is a huge challenge for universities across the UK. With such a diverse range of applications and requirements, any system must be designed to meet the varying needs of many.
For example, there have been huge advances in genome sequencing technology; the amount of data coming out of sequencers is enormous and it’s only an HPC resource that’s capable of analysing that data in a timely manner.
The procurement of a HPC system may seem to be done for the most altruistic of reasons – to enable researchers to solve some of the most difficult questions and problems that humanity faces today.
In reality though, you also need to attract highly skilled researchers to your university who can make good use of the system. Having a good-sized HPC resource attracts certain classes of researcher – who would not be able to affectively pursue their research at Universities with smaller facilities.
Thus the choice of location for research activities, groups and specialist centres is often influenced by the presence of a large and effective HPC service.
HPC is more than just a tool to enable scientific research; it’s a strategic element in attracting world-leading scientists. It attracts new postgrad researchers, entire research projects and, critically, larger grants.
Procure a machine that researchers want
To meet the needs of its researchers at the University of Bristol, the Advanced Research Computing team has a rolling programme to update its HPC facilities roughly every three years.
The team polls its researchers and scientists about what they’re investigating, how HPC can help and what they really need from a supercomputer. This information helps to inform the procurement process.
BC4 was the culmination of a lot of hard work from Bristol’s integration partner, OCF, along with technology providers DDN Storage, Lenovo and Intel. It supports researchers from across the university.
The new system uses Lenovo compute nodes with Intel Broadwell CPUs, as well as NVIDIA GPUs. The cluster is connected to one petabyte of DDN GS7K and IME storage via Intel’s Omni-path architecture. And, it’s housed at VIRTUS Data Centres in Slough, due to lack of data centre space on campus.
>See also: Giving cyber security a voice: IBM’s Watson
It’s no stretch to say that research and teaching will benefit greatly, with BC4 being one of the fastest and most advanced university supercomputing facilities in the UK. Capable of 600 trillion calculations per second, it’s the largest UK university system by core count, which puts it at number 301 on the TOP500 supercomputer list globally.
All this hard work by the technology vendors and systems integrator and the Advanced Research Computing team is with the aim of supporting the research. It’s all about squeezing out the very maximum performance from the hardware so it’s vital to work with the very best technology vendors.
Where one vendor may be great at compute, another will be much better at storage. There’d be a significant disadvantage to go with just one vendor for the whole system. This provides a good reason to work with a systems integrator that has long-standing and successful partnerships with world-leading vendors, such as Bristol’s integrator partner, OCF.
The end goal is always research
Although only launched last month, BlueCrystal 4 is playing a role in a continuing €1.8 million study into Ebola, looking at the speed of the virus evolution, and the corresponding effect on vaccines, diagnostics and treatment.
The capabilities of BC4 are invaluable to the research and will greatly speed up the analyses of the data and thus enable more timely feedback to control and manage the spread of epidemics.
>See also: Linux turns 25 years young
This enables the team, led by Dr. David Matthews, Senior Lecturer in Virology at the University, to examine how the virus had evolved over the previous year, informing public health policy in key areas such as diagnostic testing, vaccine deployment and experimental treatment options.
This complex data analysis process took around 560 days of supercomputer processing time, generating nine thousand billion letters of genetic data before reaching the virus’ 18,000 letters long genetic sequence for all 179 blood samples.
This is just one of many examples of how HPC at the University is contributing to significant research projects. Now in its 10th year of using HPC at Bristol, each phase from the first supercomputer through to BC4 has been bigger and better than the last and, in years to come this trend will definitely continue.
Sourced by Simon Burbidge, director of Advanced Computing at University of Bristol