High performance computing (HPC) is the processing of data and performing of computing calculations at high speeds. When surveyed (as part of the Sheffield Research Software survey 2020), roughly a quarter of researchers from all faculties at Sheffield expressed an interest in using HPC in the future. It may not be as difficult as you would think to do so! The barrier to entry for using specialist HPC facilities such as Graphics Processing Units (GPUs) has been significantly lowered by the emergence of software to facilitate Machine Learning (ML) and AI. The purpose of this open call is to provide funding to support skilled Research Software Engineers (RSEs) to assist researchers in migrating their research workflows to suitable HPC systems. RSE support will be provided from the Sheffield RSE group. The RSE group is a team of 12 who provide collaborative embedded support for research across the whole of the university.
Aside from the universities own HPC systems. The University of Sheffield is a partner organisation in two EPSRC Tier 2 HPC facilities; JADE II and Bede. As such we have a percentage share of the system available for University of Sheffield research. Both systems are large GPU compute clusters which have been designed to perform large scale computations in a broad range of domains. JADE II is specifically designed to support Machine Learning and Bede to support more general GPU computation spanning multiple compute nodes. All members of the university can access these facilities free of charge even if their work falls outside of EPSRCs domain.
This call will specifically aim to enable researchers to use the Tier 2 HPC systems by providing funding for embedded support within research teams or with individual researchers. Embedded support will be provided by the Research Software Engineering group, a team of ~12 software engineering experts who work collaboratively across the entire university supporting ~40 active projects worth over £12M to the university. The call is keen to support researchers from a diverse research background, particularly those who are not traditional HPC users.
There are a number themes which are support in this call;
Each of these activities is underpinned by the RSE groups ethos in good quality software engineering practice and reproducible research.
Project Title: Migrating Lung Segmentation
Application Theme: Support for scaling up of Machine Learning/AI workflows
Project Details: We worked with Bilal Tahir’s group to migrate a lung segmentation workflow to JADE. The work required the use of Niftynet a Tensor Flow based open source platform for research in medical image analysis and image-guided therapy. Niftynet uses convolutional neural networks (CNNs) and was therefore good fit to GPU acceleration on JADE. The Niftynet software was not able to be installed on JADE using a conventional approach as the installation process requires an outgoing internet connection which is unavailable with JADE for security reasons. The RSE on this project helped create a singularity container (a form of software execution environment) which packages Niftynet and all required software for use on the JADE cluster. The end result was that sears within the team could run large scale medical image analysis workflows without having complex software installations to navigate.
This call is open to all research staff but with some conditions imposed by the Tier 2 facilities. Postdocs must have at least one year left on their contract (else should apply via their PI). Doctoral research students must have support from their supervisor.
This call is specifically to facilitate uptake in the use of the Tier 2 resources. Collaborative support for use of the University HPC resources (e.g. Bessemer, ShARC) is within remit providing there is an ambition to later migrate workflows to the Tier 2 systems.
This call will fund upto £5,000 of embedded RSE time which is roughly 12 days of RSE support. There is no cash alternative available. Funding must be spent on embedded RSE time. As part of the review of your application a suitable member of the RSE team will be allocated to your project so you do not have to identify a particular team member.
Applications should not be submitted for short pieces of consultancy work which can be completed in less than a day. Instead such requests should be submitted to the Code Clinic, a free consultancy service run by the RSE and Research IT teams. Alternatively for support requests you can raise a ticket via ITS. Requests for embedded support beyond the 12 day limit will also not be considered. You should however discuss with the RSE team management about how to support your research through grant funding if a larger investment of specialist staff time is required.
If you are unsure about eligibility of scope then please contact rse@sheffield.ac.uk or book a code clinic session to discuss your project ideas.
Your application will be assessed on the following criteria;
You should apply via the provided expression of interest google form below. This is an open call and applications will be reviewed on a first-come-first-served basis.
On receiving a completed application form we will contact you to arrange a short (less than 30 minutes) online consultation to gather any additional technical information required to understand your project. The application form is purposefully short to ensure that we can discuss your requirements rather than expect you to provide all of these in the written application.
After review the projects will be ranked and estimates given as to when RSE time can be allocated. At the conclusion of your project a brief post-project report will be required.
For queries relating to collaborating with the RSE team on projects: rse@sheffield.ac.uk
Information and access to JADE II and Bede.
Join our mailing list so as to be notified when we advertise talks and workshops by subscribing to this Google Group.
Queries regarding free research computing support/guidance should be raised via our Code clinic or directed to the University IT helpdesk.