Event-based Parallel Computing
POETS is a significantly different way of approaching large, computer intensive problems.
Project leader
Dates
May 2016 to November 2021
Project staff
Prof Alex Yakovlev, Dr Ghaith Tarawneh, Allesandro de Gennaro
Sponsors
Partners
University of Southampton, Imperial College London, University of Cambridge. Chinese University of Hong Kong, e-Therapeutics Plc, Imagination Technologies Ltd UK, NMI (National Microelectronics Inst), Numerical Algorithms Group Ltd, Ricardo Group.
Description
The evolution of traditional computer technology has been exponential. We have moved from simple machines with a handful of bytes of memory and glacial clock speeds to multi-gigabyte architectures running many orders of magnitude faster. But the same fundamental process is at the heart: a central core doing one thing at a time. Over the past few years, architectures have appeared containing multiple cores. Exploiting these efficiently remains a 'holy grail' of computer science.
POETS takes an alternative approach. This is possible because of the development of cheap, small cores and massive reconfigurable platforms. A previous EPSRC project, BIMPA, enabled us to assemble a million core machine, creating a 'meta-computer'. In conventional supercomputers, we program the behaviour of each core and each communication between them explicitly. In our approach, we define a set of small, simple behaviours for the set of cores, and leave them to get on with it. With the right behavioural definitions, the system 'self-organises' to produce the desired results.
BIMPA was designed for neuroscience applications. A subsidiary research aim allowed us to study the use of the architecture for alternative (physics-based) problems. This kind of approach can lead to dramatic speed increases over conventional solution techniques.
POETS is not a general-purpose computing technique. But it is suited to a variety of traditional computer intensive engineering and research problems. It can produce results much faster than conventional machines at a fraction of the cost.
We are exploring this application arena by asking:
- what kind of architectures are best and fastest?
- how might they be automatically configured to self-organise?
- how might we build bridges between this new technology and a nascent user base?
Industry has made large investments in computing technology over the years. If POETS is to become the disruptive technology it is capable of, we need to address a serious 'hearts and minds' issue for commercial uptake to ensue.