Accelerating Scientific Discovery with Accelerated Computing

Lead Research Organisation: University of Sheffield
Department Name: Computer Science

Abstract

My proposed Fellowship will revolutionise the use of High Performance Computing (HPC) within The University of Sheffield by changing perceptions of how people utilise software and are trained and supported in writing code which scales to increasingly large computer systems. I will provide leadership by demonstrating the effectiveness of specific research software engineer roles, and by growing a team of research software engineer at The University of Sheffield in order to accommodate our expanding programme of research computing. I will achieve this by: 1) developing the FLAME and FLAME GPU software to facilitate and demonstrate the impact of Graphics Processing Unit (GPU) computing on the areas of complex systems simulation; 2) vastly extending the remit of GPUComputing@Sheffield to provide advanced training and research consultancy, and to embed specific software engineering skills for high-performance data parallel computing (with GPUs and Xeon Phis) across EPSRC-remit research areas at The University of Sheffield.

My first activity will enable long-term support of the extensive use of FLAME and FLAME GPU for EPSRC, industry and EU-funded research projects. The computational science and engineering projects supported will include those as diverse as computational economics, bioinformatics and transport simulation. Additionally, my software will provide a platform for more fundamental computer science research into complexity science, graphics and visualisation, programming languages and compilers, and software engineering.

My second activity will champion GPU computing within The University of Sheffield (and beyond to its collaborators and industrial partners). It will demonstrate how a specific area of research software engineering can be embedded into The University of Sheffield, and act as a model for further improvement in areas such as research software and data storage. I will change the way people develop

and use research software by providing training to students and researchers who can then embed GPU software engineering skills across research domains. I will also aid researchers who work on computationally demanding research by providing software engineering consultancy in areas that can benefit from GPU acceleration, such as, mobile GPU computing for robotics, deep neural network simulation for machine learning (including speech, hearing and Natural language processing) and real time signal processing.

The impact of my Fellowship will vastly expand the scale and quality of research computing at The University of Sheffield, embed skills within students and researchers (with long-term and wide-reaching results) and ensure energy-efficient use of HPC. This will promote the understanding and wider use of GPU computing within research, as well as transitioning researchers to larger regional and national HPC facilities. Ultimately my research software engineer fellowship will facilitate the delivery of excellent science whilst promoting the unique and important role of the Research Software Engineer.

Planned Impact

The UK government's Eight Great Technologies report [1] shows that the UK has the potential to demonstrate international leadership in High Performance Computing (HPC) through the development of efficient software solutions. My fellowship proposal aims to directly address the shortfall in expertise related to accelerated computing (computing which utilises dedicated co-processors to achieve increased computational performance within a tighter energy window). It will deliver via two complementary activities which are: 1) to facilitate the transition of researchers studying complex systems to HPC; 2) to deliver dedicated training and build a community for accelerated computing. My approach addresses precisely EPSRC's strategic goals to; shape capability (by developing high-quality, wide-reaching software), develop leaders (by providing bespoke training and software consultancy) and deliver impact (through meaningful industrial engagement).

The direct industrial beneficiaries of my fellowship are my project partners who have shown full commitment to the proposed programme of work. In particular, NVIDIA and Intel have a vested interest in increasing understanding of the value of their accelerated hardware because it will increase product sales and stimulate further research development. The Transport Systems Catapult (TSC) and Accelerated Computing Research Centre (ACRC) also have a vested interest in the delivery of ground-breaking and commercially useful complex systems software. The TSC remit is to provide transport simulation at national scales (with integrated modelling of transport mechanisms). This is highly dependent on the availability of novel and customised high-quality software. The TSC reports that industrial demand for products that can deliver improved simulation scale and flexibility (especially if achieved utilising accelerated computing with reduced energy implications) is high, and is not currently met by any provider. More broadly, industry partners are actively seeking to understand more about how accelerated computing can improve their products and service delivery, and thus gain a competitive advantage over their competitors (for example; DNV.GL, Google, Facebook, Baidu and Tesla Motors).

My thoughtfully developed programme of training, and my ability to lead a community that will support accelerated computing, will breed a critical mass of RSE which will add value in both academic and industry environments. Indeed, the demand for accelerated computing experts is high, yet there is no current agenda to provide this elsewhere.

[1] D. Willet, "Eight Great Technologies, Technical Report," Policy Exchange, 2013.

Publications

10 25 50
 
Description The development of the undergraduate training module has been accelerated and delivered within the first year of the fellowship. This has been a significant undertaking and has formed training material which is now being publicly disseminated and used for the delivery of shorter focused training sessions.

The FLAME development committee has been established and the code has been migrated to an open platform (Github).

A number of industrial funded projects have begun or are agreed. Each of these embed GPU computing in specific domains.
Exploitation Route The FLAME GPU software has been improved an is available for download. This will be useful to a wide range of research audiences. I am now director of a successful RSE group at the university of Sheffield which has 11 full time employees working across the university. The RSE Admin software which I have released is used for managing this group and showing that it can be sustained as cost neural to the university. For example the team has worked on ~30 research projects in 2019/2020 and recovered more than the full staff costs of the group.
Sectors Digital/Communication/Information Technologies (including Software),Education,Transport

 
Description This award has led to a number of industrial consultancy, major EU grants and smaller proof of concept projects (directly apply FLAME GPU or GPU computing). These projects have has impact on commercial software performance or delivered white paper reports on performance implications. The results of which have impact (time saved in faster simulations and better simulation tools for the evaluation of public infrastructure investment) particularly within the transport sector.
First Year Of Impact 2017
Sector Digital/Communication/Information Technologies (including Software),Transport
Impact Types Societal

 
Description Attendance of EPSRC Software Strategy Meeting
Geographic Reach National 
Policy Influence Type Membership of a guideline committee
 
Description DLI Train the Trainers, Deep Learning Fundamentals Workshop, University of Bristol
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
Impact In this full-day workshop, you will learn the basics of deep learning by training and deploying neural networks. Build the skill-set and toolbox you need to build your own deep learning solutions through hands-on projects. Learners will: Understand general terms and background of deep learning Implement common deep learning workflows such as Image Classification and Object Detection Manipulate training parameters to improve accuracy Modify internal layers of neural networks to adapt to new problems Deploy your networks to start solving real-world problems
URL http://gpucomputing.shef.ac.uk/education/2018-01-29-dli-bristol/
 
Description EPSRC Tier 2 JADE Organisational and RSE Committee Membership
Geographic Reach National 
Policy Influence Type Membership of a guideline committee
URL http://www.jade.ac.uk
 
Description Report to DfT on use of GPUs within Transport Simulation
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
 
Description Sheffield GPU Hackathon
Geographic Reach Europe 
Policy Influence Type Influenced training of practitioners or researchers
Impact Delivered a 5 day GPU hackathon to improve researchers coding skills and allow them to target new HPC architectures as part of their research. The Hackathon specifically targeted nationally leading software infrastructure.
URL https://gpuhack.shef.ac.uk/2019/index.html
 
Description Talk and Diversity Co-Chair, Conference of Research Software Engineers UKRSE 2017
Geographic Reach Multiple continents/international 
Policy Influence Type Influenced training of practitioners or researchers
Impact The conference featured a mix of talks, posters, discussion workshops and hands-on tutorials. All together these will provided attendees with the opportunity to learn new skills, discuss the state of research software engineering, network and share knowledge.
URL http://rse.ac.uk/conf2017/
 
Description Training MPs on Deep Learnng and AI at House of Commons
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
URL https://blogs.nvidia.com/blog/2018/02/28/ai-workshop-uk-lawmakers/
 
Description Voted Trustee and Elected Vice President of the RSE Society
Geographic Reach National 
Policy Influence Type Membership of a guideline committee
Impact Paul Richmond and Mozhgan Kabiri Chimeh both voted as trustees for the RSE Sicty. Paul Richmond voted and elected as the Vice President. The Society of Research Software Engineering was founded on the belief that a world which relies on software must recognise the people who develop it. Our mission is to establish a research environment that recognises the vital role of software in research. We work to increase software skills across everyone in research, to promote collaboration between researchers and software experts, and to support the creation of an academic career path for Research Software Engineers.
URL https://society-rse.org/
 
Description A Centre of Excellence in Computational Biomedicine (CompBioMed)
Amount £365,095 (GBP)
Funding ID 142710 
Organisation European Commission H2020 
Sector Public
Country Belgium
Start 10/2015 
End 09/2018
 
Description Accelerating Scientific Discovery with Accelerated Computing
Amount £721,298 (GBP)
Funding ID 145175 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 02/2016 
End 01/2021
 
Description CompBioMed2
Amount £321,162 (GBP)
Funding ID 156659 
Organisation European Commission H2020 
Sector Public
Country Belgium
Start 10/2019 
End 09/2023
 
Description Confidence in Concept
Amount £106,556 (GBP)
Funding ID 152050 
Organisation Medical Research Council (MRC) 
Sector Public
Country United Kingdom
Start 08/2017 
End 07/2018
 
Description Digital Fly Brain
Amount £530,873 (GBP)
Funding ID 142049 
Organisation Biotechnology and Biological Sciences Research Council (BBSRC) 
Sector Public
Country United Kingdom
Start 05/2015 
End 04/2018
 
Description EPSRC IAA COLLABORATIVE R&D
Amount £25,000 (GBP)
Funding ID 148541 
Organisation Siemens AG 
Sector Private
Country Germany
Start 07/2016 
End 01/2017
 
Description Mechanisms of integrating growth and patterning during development by the atypical cadherins Fat and Dachsous
Amount £481,262 (GBP)
Funding ID 153345 
Organisation Biotechnology and Biological Sciences Research Council (BBSRC) 
Sector Public
Country United Kingdom
Start 10/2018 
End 09/2021
 
Description Optimisation - extra allocation
Amount £6,500 (GBP)
Funding ID 148957 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 06/2016 
End 06/2016
 
Description Other (Atkins)
Amount £17,500 (GBP)
Funding ID 149355 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 03/2017 
End 05/2017
 
Description Other (Atkins)
Amount £5,000 (GBP)
Funding ID 149355 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 03/2018 
End 03/2018
 
Description Other (Atkins)
Amount £12,500 (GBP)
Funding ID 149355 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 09/2017 
End 11/2017
 
Description Other (Atkins)
Amount £20,000 (GBP)
Funding ID 149355 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 05/2017 
End 08/2017
 
Description Other (Atkins)
Amount £7,500 (GBP)
Funding ID 149355 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 12/2017 
End 02/2018
 
Description Other -Atkins
Amount £45,000 (GBP)
Funding ID 145719 
Organisation Transport Systems Catapult 
Sector Private
Country United Kingdom
Start 10/2016 
End 03/2017
 
Description Other DfT
Amount £22,000 (GBP)
Funding ID 149484 
Organisation Department of Transport 
Sector Public
Country United Kingdom
Start 09/2016 
End 01/2017
 
Description RIA
Amount € 867,273 (EUR)
Funding ID 152008 
Organisation European Commission 
Department Horizon 2020
Sector Public
Country European Union (EU)
Start 02/2018 
End 12/2022
 
Description RRUK - Rail Research UK Grant
Amount £124,324 (GBP)
Organisation Rail Safety and Standards Board 
Sector Public
Country United Kingdom
Start 01/2017 
End 03/2018
 
Description RSSB vs KE
Amount £89,140 (GBP)
Funding ID 163040 
Organisation NIHR/HEFCE Higher Education Fund for England 
Sector Public
Country United Kingdom
Start 11/2019 
End 01/2021
 
Description SATGPU - Phase III
Amount £12,500 (GBP)
Funding ID 154555 
Organisation WS Atkins 
Sector Private
Country United Kingdom
Start 09/2017 
End 11/2017
 
Description SATGPU - Phase IV
Amount £7,500 (GBP)
Funding ID 155823 
Organisation WS Atkins 
Sector Private
Country United Kingdom
Start 12/2017 
End 02/2018
 
Description SATGPU - Phase V
Amount £5,000 (GBP)
Funding ID 156516 
Organisation WS Atkins 
Sector Private
Country United Kingdom
Start 03/2018 
End 03/2018
 
Description SATGPU Phase VI
Amount £5,000 (GBP)
Funding ID 157651 
Organisation WS Atkins 
Sector Private
Country United Kingdom
Start 04/2018 
End 08/2018
 
Description Siemens - Commercial
Amount £68,000 (GBP)
Funding ID 151328 
Organisation Siemens AG 
Department Siemens plc, Keele
Sector Private
Country United Kingdom
Start 02/2017 
End 01/2018
 
Description Computational Chemistry 
Organisation University of Sheffield
Department Automatic Control and Systems Engineering
Country United Kingdom 
Sector Academic/University 
PI Contribution Proof of concept work to demonstrate the impact of GPU computing in the domain of Particle balance equations and Monte-Carlo methods. This collaboration is ongoing, a working document has been prepared and we expect to generate either a journal paper or funding application.
Collaborator Contribution This partnership is a collaborative research software project. The partners involved bring their domain science and myself and my team bring novel computational methods.
Impact A working paper has been developed (for the purposes of communicating collaborative ideas between partners rather than for submission to a journal/conference). A piece of software has been developed which reproduces existing results but uses new GPU technology (https://github.com/FLAMEGPU/FLAMEGPU/tree/monteCarlo_dev)
Start Year 2016
 
Description How do the atypical cadherins Fat and Dachsous integrate growth and patterning during development? 
Organisation University of Sheffield
Department School of Mathematics and Statistics
Country United Kingdom 
Sector Academic/University 
PI Contribution This project is under submission with BBSRC. It is a collaboration in which my Research Software Engineering team will provide software engineering support for 6M of software development in multi agent systems software development.
Collaborator Contribution The partners are focusing on the underlying biological research questions where as my team is delivering research software support to assist with this.
Impact BBSRC full proposal currently under review
Start Year 2017
 
Description Insigneo Research Software Engineering Support 
Organisation University of Sheffield
Country United Kingdom 
Sector Academic/University 
PI Contribution My RSE team is providing research support through two members of staff (100% FTE and a 60% FTE). This covers research software consultancy and HPC support.
Collaborator Contribution The partner is a large research institute with a focus on multi-scale modelling
Impact Delivery of HPC training and support, Delivery of software engineering on open source image registration software
Start Year 2017
 
Description Rail and Transport Simulation 
Organisation University of Sheffield
Department Department of Infection, Immunity and Cardiovascular Disease
Country United Kingdom 
Sector Academic/University 
PI Contribution Application of GPU technology to the simulation and optimisation of Rail and multi-model transport systems.
Collaborator Contribution Provision of rail network models within the wider context of simulating passengers end to end journeys.
Impact Grant award from RRUK Two grant awards from SIEMENS Rail automation Dissemination video (https://www.youtube.com/watch?v=Rz_XzbZIMes)
Start Year 2016
 
Title FLAME GPU Software updates 
Description Continued releases of Software updates for the FLAME GPU software package. FLAME GPU is a multi agent simulator which targets GPU architectures. 
Type Of Technology Software 
Year Produced 2017 
Open Source License? Yes  
Impact New functionality being described in papers currently under development. 
URL http://www.flamegpu.com
 
Title GPU Computing and Sheffield Website 
Description A website for the GPU Computing community at sheffield. 
Type Of Technology Webtool/Application 
Year Produced 2017 
Impact Availability of on demand training. Hub for a community of researchers and developers. 
URL http://gpucomputing.shef.ac.uk
 
Title JADE and JADE support wesbite 
Description Website and automated documentation site for JADE Tier 2 facility 
Type Of Technology Webtool/Application 
Year Produced 2017 
Open Source License? Yes  
Impact User support and information for national JADE users. 
URL http://www.jade.ac.uk
 
Title Osteolytica 
Description Osteolytica is a software tool for osteolytic lesion analysis developed at the University of Sheffield. It specialises in the reconstruction and analysis of volumetric bone samples generated by murine myeloma studies and obtained from Micro CT scanners within a research environment. 
Type Of Technology Software 
Year Produced 2017 
Impact Funding from TUOS Impact fund to continue development for 
URL http://www.osteolytica.com/
 
Title RSEAdmin 
Description This is a tool for tracking grant applications, managing RSE commitment and reporting on staff expenses and cost recovery. The site has two users types (Admin and RSE) which have different levels of permissions. RSE users can view, edit and create projects as well as see allocations to projects based on effort (i.e. days). Admin users are able to create allocations on projects (allocated by effort or budget) and view reports on group finance and staffing. There is a distinction between Allocated Projects (i.e. Directly incurred projects) where staff generate overheads for the university and Service Projects which are effort based and generate overheads for the group or facility. 
Type Of Technology Webtool/Application 
Year Produced 2019 
Open Source License? Yes  
Impact Software is used for managing the RSE group at Sheffield which is run by Paul Richmond. Software is being considered for use by a number of other institutions. 
URL https://rseadmin.readthedocs.io/en/latest/
 
Title Research software Engineering Website 
Description New website for the RSE community at Sheffield 
Type Of Technology Webtool/Application 
Year Produced 2016 
Impact Availability of on demand training. Central contact point for the emerging RSE groups (co-led with Mike Croucher). 
URL http://www.rse.shef.ac.uk
 
Title SATGPU 
Description GPU accelerated version of the SATURN transport simulation software. To be released in March 2018 under licence from Atkins. Announced at the 2017 SATURN AGM. 
Type Of Technology Software 
Year Produced 2018 
Impact Software is currently being beta tested by Highways agency and TfL. Adoption of the software will lead to a national step change in the use of GPUs for transport simulation within transportation consultants. 
URL https://saturnsoftware2.co.uk/uploads/files/SATUGM17-SATGPU2.pdf
 
Title Website for agent based modelling group 
Description Website for interest group in agent based modelling 
Type Of Technology Webtool/Application 
Year Produced 2017 
Open Source License? Yes  
Impact Workshops, seminars and meetings relating to agent based modelling at TUOS. 
URL http://agentbasedmodels.group.shef.ac.uk
 
Description A talk or presentation - Participation in an activity, workshop or similar - Singularity on the ShaRC 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact A colleague did a presentation /talk on Singularity on the ShARC presentation, University of Sheffield, 29th August 2017
Year(s) Of Engagement Activity 2017
 
Description 2day CUDA Training Event at Imperial College London 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Assistance in the delivery of an NVIDIA sponsored 2 day training event for GPU computing.
Year(s) Of Engagement Activity 2016
 
Description An RSE from UK 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited Lightning Talk: "An RSE from UK". Software Engineering for Computational Science and Engineering BoF session at SC High Performance Conference, Dallas, Texas. November 2018.
Year(s) Of Engagement Activity 2018
URL https://betterscientificsoftware.github.io/swe-cse-bof/2018-11-sc18-bof/
 
Description Best Practices for HPC Training and Education workshop 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited as a panellist for Best Practices for HPC Training and Education workshop - Scheduled for SC High Performance Conference. November 2019.
Year(s) Of Engagement Activity 2019
URL https://sc19.supercomputing.org/proceedings/workshops/workshop_pages/wksp140.html
 
Description Building Resilience in Academia 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited Talk: "Building Resilience in Academia". Sheffield Computer Science Society at the University of Sheffield, Sheffield, UK. November 2018.
Year(s) Of Engagement Activity 2018
 
Description CUDA GPU training workshop Dundee 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact CUDA and deep learning training delivered at Dundee University on behalf of NVIDIA.
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/dundee/
 
Description CUDA and Deep Learning Training at the University of Glasgow 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Other audiences
Results and Impact 3 day CUDA and Deep Learning training provided at the University of Glasgow on behalf on NVIDIA
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/cuda-glasgow/
 
Description DLI Ambassador Deep Learning Fundamentals Workshop, University of Sheffield 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact In this full-day workshop, attendees learned the basics of deep learning by training and deploying neural networks.
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/dliworkshop/
 
Description Delivery of Parallel Compuitng with GPUs undergraduate and postgraduate course Spring 2016, 2017 and 2018 Semester 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Undergraduate students
Results and Impact Delivery of new teaching module on GPU computing. Course content made publicly available via NVIDIA educational resources.
Year(s) Of Engagement Activity 2016,2017,2018
URL http://paulrichmond.shef.ac.uk/teaching/COM4521/index_2016/
 
Description FLAME GPU Tutorial at ALIFE 2019 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Paul and Mozhgan Organized a (2Hr) tutorial at The 2019 Conference on Artificial Life (ALIFE 2019), Newcastle, UK (Simulating Complex Systems with FLAME GPU) - July 2019
Year(s) Of Engagement Activity 2019
URL https://2019.alife.org/programme/workshops-tutorials/
 
Description FLAME GPU: Complex System Simulation Framework 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Panelist Women in HPC (WHPC) workshop at SC High Performance Conference, Dallas, Texas. November 2018.
Year(s) Of Engagement Activity 2018
URL https://womeninhpc.org/events/sc18/workshop
 
Description FLAME GPU: Complex System Simulation Framework workshop at HPCS 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact FLAME GPU: Complex System Simulation Framework (4Hr). Genoa, Italy July 2017, The 2017 International Conference on High Performance Computing and Simulation (HPCS 2017)
Year(s) Of Engagement Activity 2017
URL http://hpcs2017.cisedu.info/conference/workshops---hpcs2017/workshop11-shpcs
 
Description Hosted Two Day Intel Parallel Computing Training 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Two day course for research community at the University of Sheffield.
Year(s) Of Engagement Activity 2017
URL http://rse.shef.ac.uk/training/intel/
 
Description Hosted Two Day Training from NVIDIA for Deep Learning Education 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Two days of training hosted for Sheffield University Research Community.
Year(s) Of Engagement Activity 2016
 
Description InstanceHub: Simple provisioning of cloud instances for training 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Demo talk at UKRI Cloud workshop
Year(s) Of Engagement Activity 2020
URL https://cloud.ac.uk/workshops/mar2020/instancehub-simple-provisioning-of-cloud-instances-for-trainin...
 
Description Introduction to CUDA workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact One day intorduction to CUDA course
Year(s) Of Engagement Activity 2018
URL http://gpucomputing.shef.ac.uk/education/sheffield_onedaycuda/
 
Description Introduction to Deep Learning on DGX-1 workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact Local training course on Deep Learning provided by my RSE team.
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/intro_dl_sharc_dgx1/
 
Description Invited Mentor at GPU Hackathon Lugano 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Paul Richmond Invited Mentor for GPU Hackathon in Lugano Switzerland.
Year(s) Of Engagement Activity 2019
URL https://www.cscs.ch/publications/news/2019/eurohack19-gpu-programming-hackathon/
 
Description Invited Talk - University of Glasgow 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Professional Practitioners
Results and Impact Invited talk on HPC
Year(s) Of Engagement Activity 2020
 
Description Large Scale Multi-Agent Simulation using FLAME GPU 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Policymakers/politicians
Results and Impact Mozhgan Invited Talk and Demonstration: "Large Scale Multi-Agent Simulation using FLAME GPU", HPCS (Leading international conference in HPC) - 2019
Year(s) Of Engagement Activity 2019
 
Description Mentor at Helholtz GPU Hackathon 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited mentor by Nvidia - Helmholtz GPU Hackathon, Germany (April 2019) after Peer reviewed submitted application
Year(s) Of Engagement Activity 2019
 
Description Participation in an activity, workshop or similar - ARCHER workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact A colleague presented at the ARCHER Workshop on the current state of JADE, Hartree, Daresbury 27th June 2017
Year(s) Of Engagement Activity 2017
 
Description Participation in an activity, workshop or similar - Containers for HPC workshop 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact A colleague presented at Containers for HPC, Cambridge University, 29th and 30th June 2017 - Title : Singularity for GPU and Deep Learning
Year(s) Of Engagement Activity 2017
URL https://rse-cambridge.github.io/hpc-container-workshop/
 
Description Program Co-Chair (co-organizer), Workshop/Conference of Agent Based Modelling ABM 2017 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact This mini-conference showcased a wide range of topics in agent-based modelling, bringing together leading researchers from the Universities of Sheffield, Leeds, and York to discuss latest advances in the simulation of complex systems. The event brought together academics and researchers from various disciplines, encouraging discussion and a lively debate. Through this event, our aim was to grow collaborations that would have a substantial contribution to future in silico research.
Year(s) Of Engagement Activity 2017
URL http://agentbasedmodels.group.shef.ac.uk/2017/10/31/agent-based-modelling-abm-2017-mini-conference/
 
Description RSEAdmin a project management tool for RSE group management 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Talk at the RSE Leaders meeting in Edinburgh December 2019
Year(s) Of Engagement Activity 2019
 
Description Simulating Complex Systems with FLAME GPU Workshop 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Simulating Complex Systems with FLAME GPU (2Hr). Lyon, France September 2017, 14th European Conference on Artificial Life (ECAL 2017)
Year(s) Of Engagement Activity 2017
URL https://project.inria.fr/ecal2017/tutorials-2/
 
Description Software Engineering for Computational Science and Engineering 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited Talk: Software Engineering for Computational Science and Engineering BoF session at ISC High Performance Conference, Frankfurt, Germany. June 2019
Year(s) Of Engagement Activity 2019
 
Description Talk/presentation at ABM 2017 mini conference : From biological cells to populations of individuals: Complex Systems Simulations with CUDA 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Complex systems are prevalent throughout various levels of biology from the molecular and cellular scales through to populations of interacting individuals. This talk discusses how formal state based representation of agents within a complex system can be simulated and visualised at large scales using the open source FLAME GPU framework. Methods of code generation from XML documents and use of CUDA streams for heterogeneous state execution are presented. Examples include cellular tissue modelling and large scale crowd dynamics.

Agent Based Modelling (ABM) is a powerful simulation technique used to assess and predict group behaviour from a number of simple interacting rules between communicating autonomous agents. Traditional ABM toolkits are primarily aimed at a single CPU architecture, with an inherent lack of parallelism and a reliance on deeply serialised algorithms which are not well suited to many core architectures. As a result, the uptake of GPUs has had a relatively low impact on the field of ABM. The few examples of ABM on the GPU are limited to specific implementations of discrete space (cell) models or swarm systems.

The Flexible Large-scale Agent Modelling Environment for the GPU (FLAME GPU) addresses the performance and architecture limitations of previous work by presenting a flexible framework approach to ABM on the GPU. Most importantly it addresses the issue of agent heterogeneity through the use of a formal state based agent representation. This representation allows agents to be separated into associated state lists which are processed in batches to allow very diverse population of agents whilst avoiding large divergence in parallel code kernels.

In previous work FLAME GPU has been used in a wide range of research applications from the simulation of cellular level systems biology models of wound formation to pedestrian simulation. The overall aim of FLAME GPU is to provide a technique that allows modellers to harness parallelism and the power of the GPU, without requiring background knowledge of parallelism or the hardware (a domain specific modelling tool). Performance rates equalling, or bettering that of HPC clusters can easily be achieved, with obvious cost to performance benefits. Massive population sizes can be simulated, far exceeding those that can be computed (in reasonable time constraints) within traditional ABM toolkits. Finally the use of the GPU allows simulations to be visualised in real time, allowing interaction and real time steering.

FLAME GPU is currently undergoing a major update which includes the addition of the use of atomics to improve sorting performance for building spatial data structures, CUDA streams to permit simultaneous execution of differing agent populations (or state configurations) and a runtime API to simplify code generation and manage simultaneous simulations (in order to explore sets of parameters). The expected results of the following updates are that agents with diverse states should receive vastly improved performance over the already impressive results compared to existing ABM simulators.

In summary this talk will give an overview of the FLAME GPU approach to state based agent simulation, discuss agent communication algorithms and recent improvements using atomics, present methods for code generation from XML model descriptions and give late breaking performance results for simulations of cellular populations and crowds of pedestrians each with over a million agents.
Year(s) Of Engagement Activity 2017
URL http://agentbasedmodels.group.shef.ac.uk/2017/10/31/agent-based-modelling-abm-2017-mini-conference/
 
Description The Art of Claiming Credit 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Mozhgan Invited Talk: "The Art of Claiming Credit". Women in HPC (WHPC) workshop at ISC High Performance Conference, Frankfurt, Germany. June 2019
Year(s) Of Engagement Activity 2019
 
Description Two Day CUDA Training Course 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact Delivery of a two day introduction to GPU Computing course at the University of Sheffield.
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/cuda/
 
Description Two Day GPU Training as part of "Deep Learning Big Data and Big Compute" 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Delivery of two days of training and talks.
Year(s) Of Engagement Activity 2016
URL https://dlwensias.wordpress.com/home/
 
Description Two day introduction to Deep Learning with GPUs Course 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other audiences
Results and Impact Two day course delivered by a member of my staff (part funded by the grant) for researchers and academics.
Year(s) Of Engagement Activity 2017
URL http://gpucomputing.shef.ac.uk/education/intro_dl_sharc_dgx1/