Manycore Research Innovation and Opportunities Network (MaRIONet)

The transition to manycore systems is arguably the greatest challenge
facing the ICT industry, Computational Science, and Computing Science
research. The challenge is globally recognised, and is an EPSRC ICT priority:
Manycore Architectures in Distributed and Embedded Systems (MACDES).
The proposed network will identify and bring together a critical mass
of UK manycore researchers.

The manycore revolution is fundamentally changing multiple levels of
the computation execution stack: from processor architecture, through
systems software, to applications. The UK has world leading
researchers and industrialists working on specific technologies at
each level, but currently researchers and engineers do not collaborate
adequately on the complex manycore challenges. Furthermore,
meaningful industry engagement is limited, despite the large number of
UK-based industrial research labs with strategic interest in manycore
systems.

We will foster effective communication between academic and industrial
research teams both at different levels in the execution stack
(e.g. compiler developers targeting an emergent architecture), and
within each level (e.g. program analysis applied to improve memory
access latency in the systems layer). The teams will benefit by being
able to identify and exploit synergies and opportunities for
cross-pollination.

We will establish well defined, highly available channels for
communication and low-overhead mechanisms for collaboration. Given
that many world-class UK researchers are isolated in relatively small
departments and research groups, the network will enable national
sharing of ideas, information and infrastructure. Community support
and resource pooling will enable UK researchers to achieve greater
academic impact globally, competing effectively with larger,
well-resourced groups in the USA (e.g. Aspire Lab at Berkeley), China
(e.g. State Key Lab of Computer System and Architecture, Chinese
Academy of Sciences), and elsewhere.

Both architectural innovation and systems software development are
massive investments, often involving person-decades of
effort. MaRIONet will enable sharing of expertise and infrastructure
to lower the barrier to entry for UK systems researchers working in
this area. Industrial network partners will gain exposure to
cutting-edge research, and ideas to address current problems. Academic
partners gain real-world use cases and grand challenges.

At least three of the UK Policy Exchange `Eight great technologies'
are directly enabled by the increased scale and speed of future manycore
systems and software, namely:

* Big Data revolution: enabled by manycore systems for
energy-efficient collection, analysis and storage of big data.

* Robotics and autonomous systems: enabled by manycore
systems for artificial intelligence, planning, image recognition,
humanoid motion control.

* Synthetic biology: enabled by manycore processing of biological
data, in-silico simulation of bio-mechanisms.

Hence we anticipate that future UK economic growth and societal well-being
will be strongly influenced by manycore technology, which is the focus of
our research network.

We envisage impact primarily in the areas of Knowledge, People and
Economy. We denote RCUK impact activities as: Application and
exploitation(APP), Communication & engagement (COMM), Collaboration &
co-production(COLL) and Capacity & involvement(CAP).

KNOWLEDGE

MaRIONet enhances the capability of the fragmented UK manycore research
community by creating a critical mass, to deliver better representation
of UK-based research in leading international conferences, (e.g. PLDI,
ASPLOS), journals (e.g. IEEE TOCS, ACM TOPLAS), and events
(e.g. HiPEAC, UK-MAC). (COMM,CAP)

Direct contributions to knowledge include open access reports and
lecture notes. A key contribution is the roadmap that should inform
future funding priorities. (COMM,CAP)

The impact of new technologies developed within the network will be
enhanced by ongoing engagement with industrial network partners. We
will foster collaborations, including academic/industrial
collaborations, via regular network meetings, industrial fora like
EdLambda, and by Innovate UK KTN events. (APP,CAP)

Open access to documents and software will facilitate
dissemination. Similarly open access to software and our infrastructure
sharing scheme will encourage technology adoption and multi-technology
evaluation by making cross-site experimentation easier. (APP,COLL)

PEOPLE

We will facilitate the development of new skills for current
practitioners, and the training of new practitioners as part of the UK
people pipeline. Moreover we seek to engage the wider public with social
media, and school children via classroom materials. (CAP)

Postgraduate students and early career researchers will enhance both
discipline-specific competence and transferable skills by attending
themed meetings, annual workshops and the Summer School. The large
postgraduate and early career community identified by the network will
be a key part of a nationwide virtual centre of excellence for
manycore. An early career mentoring champion on the steering committee
will provide informal advice and career guidance at annual
workshops. Industrial partners will provide placements/internships for
early career researchers. (CAP, COLL)

Our social media strategy exploits channels like twitter and Youtube to
present network activities to a wider audience, e.g. three-minute videos
summarising new research. We will raise the exposure and clarify the
motivation of our publicly-funded research. (COMM)

Our schools outreach strategy will exploit existing strong links with
the Computing At School grassroots ICT teachers organization. The PI
will design manycore classroom activity packs to support computational
thinking. The packs will be developed by interns at Glasgow, then
distributed via Computing At School, and using online media
channels. (CAP, COMM)

ECONOMY

Effectively exploiting manycore architectures is arguably the greatest
challenge facing the UK ICT sector, which directly employs 1m people and
is worth GBP 58 billion annually. MaRIONet seeks to strengthen the
manycore capabilities of the sector and secure them as a core competency
for UK ICT.

We will exploit strong industrial links of the initial network partners,
and target strong growth in industrial partners during the project.
Relevant technical challenges from industrial partners at network
meetings will foster academic/industrial collaborative projects. These
may start as hackathons or internships, and then transition into more
substantial collaborations. Some may leverage Knowledge Transfer Account
(KTA), or other, funding. (COLL, CAP)

We will encourage industrial exploitation and application of academic
research. Individual members have access to institutional help, e.g.
Easy Access IP. A knowledge transfer champion on the steering committee
will provide expert consultancy to academic and industrial network
partners about best practice for research exploitation. (COLL, CAP)