Many-core architectures and concurrency in distributed and embedded systems
What was the priority?
With single core processors at or near their maximum clock speed, it is widely recognised that novel architectures are required. These should deliver increases in computational power within a restricted energy envelope; necessary to enable future electronic devices.
There has already been a shift by industry to multiple cores but there are aspirations to move further, to many cores. A wealth of opportunities exists if an architecture containing hundreds or thousands of cores can be properly exploited, but massive parallelism raises challenges in both the hardware and software research areas. Recognised globally as a key topic, the UK can contribute most effectively to the larger parallel agenda through its inherent strength in embedded systems research. Computation shared over multiple elements also presents particular challenges in distributed systems; another UK strength.
Implementation of the Priority
In March 2012 researchers from across the ICT portfolio attended a workshop to identify challenges which lie underneath this priority and help inform the evolution of the EPSRC strategy over the 2011-2015 Delivery Plan. The outputs of the workshop were published as a report in April 2012 - Many-core Architectures and Concurrency in Distributed and Embedded Systems (MACDES) workshop report (PDF 1.3MB).
The SADEA: System Approaches to Distributed and Embedded Architectures call launched in June 2012 taking into consideration the workshop outcomes. Proposals were invited from across the ICT landscape which addressed this priority. The minimum requirement of this call was that the research enabled cross-layer or end-to-end system considerations.
Alongside specific funding activities, the ICT community are also encouraged to submit proposals contributing to the MACDES priority using our other mechanisms, in particular the Programme Grants and Fellowships schemes.
Outcomes of the Priority
There are now a number of key groups and individuals in the UK working towards the goals of the MACDES priority and taking a cross-layer or end-to-end system approach. Some of these are supported through the proposals funded as a result of the SADEA call, alongside a substantial portfolio of programme grants, fellowships and standard mode grants.
The programme grants with most relevance to this priority are:
- REMS: Rigorous Engineering for Mainstream Systems (GoW EP/K008528/1), University of Cambridge, Professor Sewell;
- PRiME: Power-efficient, Reliable, Many-core Embedded systems (GoW EP/K034448/1), University of Southampton, Professor Al-Hashimi;
- PAMELA: a Panoramic Approach to the Many-CorE LAndscape - from end-user to end-device: a holistic game-changing approach (GoW EP/K008730/1), The University of Manchester, Professor Furber;
- From Data Types to Session Types---A Basis for Concurrency and Distribution (GoW EP/K034413/1), University of Edinburgh, Professor Wadler.
The Manycore Research Innovation and Opportunities Network (MaRIONet) launched in November 2016 and aims to bring together this critical mass of researchers to encourage further academic and industrial collaboration across the complex manycore landscape.