Queen Mary University of London (QMUL) and Newcastle University, are the first UK academic organisations to deploy servers containing IBM’s POWER9 system.
Working with OCF, a HPC integrator, both Universities have taken delivery of the systems that will be integrated by OCF into the universities existing HPC infrastructure.
‘QMUL is a world-leader in deep learning research in post-event video forensics and analysis. Deep learning by large-scale convolutional neural networks – a category of neural networks that has been proven effective in areas such as image recognition and classification – has radically changed research into computer vision in recent years,’ comments Professor Sean Gong, Head of the Computer Vision Research Group at QMUL. ‘Given our previous test trials on IBM Minsky POWER8 servers, we expect to see significant benefit from the new POWER9 servers for deep learning on big video data.’
QMUL needs the latest computing technology to enable its research using deep learning to solve some of today’s toughest scientific challenges. The University became one of the first organisations to purchase two IBM Accelerated Compute Servers (AC922) powered by POWER9 CPUs, Volta GPUs and featuring the NVLink 2.0 interconnect.
Within the Mathematics, Statistics and Physics departments at Newcastle University, researchers were looking to use the GPU-accelerated platform for computational physics projects. ‘The major difficulty we face is that our calculations are non-linear, time-dependent and three-dimensional, so solving them is out of reach with pencil and paper, and the numerical computation requires large memory and fast speed - we are humbled from the start,’ commented Carlo Barenghi, co-director of the Joint Quantum Centre at Newcastle University. ‘We did some investigations and the IBM POWER9 system was the best technology for our work – in trial runs we got a “speed up factor” in the order of 10x magnitude, so the decision was easily made.’
The AC922 leverages IBM’s new POWER9 processor with modern connectivity capabilities, which improves data movement by up to 5.6x over the PCIe Gen 3 buses found in x86 servers. IBM Power Systems deliver the only architecture enabling NVLink between CPUs and GPUs, unlocking new potential to optimise accelerated computing applications.
‘Modern AI, HPC and Analytics workloads are driving an ever-growing set of data intensive challenges,’ comments Julian Fielden, Managing Director of OCF. ‘These challenges can only be met with accelerated infrastructure, such as IBM’s POWER9. In such a highly competitive field as academic research, providing superior HPC services to compute large quantities of data quickly, can help to attract world-class researchers, as well as grants and funding.’
Thomas King, Assistant Director for Research in IT Services at Queen Mary University of London, comments: ‘Working with IBM, OCF, and Univa, our resource scheduling partner, we're supporting a big-uptake in research into Artificial Intelligence and this is a great demonstration of how the University is investing in cutting-edge research equipment for researchers and students.’
