HPC Research Accepted for Publication at the ICS 2022 Conference
Two papers by researchers at the Department of Computer Science have been accepted to the 36th ACM International Conference on Supercomputing ICS 2022 to be held on the 28-30th of June this year. ICS is one of the most prominent and revered conferences in High Performance Computing, highly regarded by the HPC community for publishing leading-edge research in this area. The two papers accepted are:
- High Throughput Multidimensional Tridiagonal System Solvers on FPGAs (Preprint) by Kamalavasan Kamalakkannan and Gihan Mudalige at Warwick, together with Istvan Reguly (PPCU) and Suhaib Fahmy (KAUST).
- Clairvoyant: A Log-Based Transformer-Decoder for Failure Prediction in Large-Scale Systems by Khalid Alharthi and Arshad Jhumka at Warwick, together with Sheng Di, Franck Cappello at Argonne National Laboratory. Preprint. The ACM ICS2022 full program can be found here.
Winner of the Faculty of Science, Engineering and Medicine Post-Doctoral Research Prize 2022
Gunduz Vehbi Demirci has been awarded with the Faculty of Science, Engineering and Medicine Post-Doctoral Research Prize 2022 for his paper jointly with Prof. Hakan Ferhatosmanoglu, "Partitioning sparse deep neural networks for scalable training and inference", published in the Proceedings of the ACM International Conference on Supercomputing (ICS '21) (DOI: https://doi.org/10.1145/3447818.3460372).
Training large-scale deep learning models is notoriously difficult. Gunduz develops a highly parallel solution to scale training of sparse deep learning models, which is combined with a novel combinatorial optimisation built on a hypergraph partitioning model, reducing parallelisation overheads and achieving computational balance among processors. An end-to-end software solution is released, enabling competing with big tech companies that have access to large infrastructures and datasets.
The work is summarised in a paper accepted by the 2021 ACM International Conference on Supercomputing, which is a premier conference in high-end systems. The research output will have a great potential to bring significant practical impact in long term as developing such comprehensive solutions takes time and is typically achieved only within large groups.
Members of the High-Performance and Scientific Computing Group (HPSC) at the department of Computer Science has won a best paper award at the 28th IEEE International Conference on High-Performance Computing, Data and Analytics held on the 17th-18th of December. The winning paper titled Predictive Analysis of Large-Scale Coupled CFD Simulations with the CPX Mini-App, develops a novel representative (mini-)application, specifically designed to model coupled execution of multi-physics numerical simulation codes from the CFD domain. The mini-coupler, CPX, is the first of its kind, combining multiple CFD mini-app instances to predict the run-time and scaling behaviour of large scale coupled CFD simulations, on modern multi-core and many-core clusters such as used for production turbomachinery design at Rolls-Royce plc. The work was carried out by PhD candidate, Archie Powell, in collaboration with Kabir Choudry, Arun Prabhakar, and Gihan Mudalige at the Department of CS Warwick, Dario Amirante (University of Surrey), Istvan Reguly (PPCU) and Stephen Jarvis (University of Birmingham).
The work was funded by the EPSRC Prosperity Partnership in Computational Science for Advanced Simulation and Modelling of Engineering Systems (AsiMoV) and Rolls-Royce plc.
Dr. Gihan Mudalige at the University of Warwick’s Department of Computer Science have been awarded an Engineering and Physical Sciences Research Council (EPSRC) ExCALIBUR research grant as part of a consortium of researchers including the Science and Technologies Facilities Council (STFC), universities of Warwick, Newcastle, Cambridge, Southampton and led by Imperial College London.
This 3 year, £2.6M project brings together communities from the UK Turbulence Consortium (UKTC) and the UK Consortium on Turbulent Reacting Flows (UKCRF) to ensure a smooth transition to exascale computing, with the aim to develop transformative techniques for advancing their production simulation software ecosystems dedicated to the study of turbulent flows. It is part of the ExCALIBUR (Exascale Computing ALgorithms and Infrastructures Benefiting UK Research) programme, aimed at delivering the next generation of high-performance simulation software for the highest-priority fields in UK research.
We are happy to announce that Dr Theo Damoulas has been promoted to Professor, effective from 1 August 2021. Quoting from Theo's recommendation,
Since his appointment to Warwick in 2015, Dr Damoulas has steadily and speedily risen to one of the most successful researchers in his two departments of Computer Science and Statistics. ... An impressive characteristic of Dr Damoulas’s expertise is his bridging of fundamentals and applications in these fields, and his ability to collaborate intensively over a wide area spanning several Warwick departments, several groups at the Turing, government bodies and industrial partners. ... Dr Damoulas has been one of the most energetic and most appreciated teachers in the department. ... The teaching achievements of Dr Damoulas extend also beyond Warwick, including his work with NYPD, his mentorships within the Turing PhD enrichment scheme, and his PhD external examining work.
it remains to say many congratulations!
Each year, the University of Warwick's Faculty of Science, Engineering and Medicine (SEM) funds a prize for the best Warwick-affiliated research output from an Early Career Researcher. Each department nominates a winner, or joint winners, out of the applications received after a judging process as determined by the Faculty.
Professor Yulan He, who led the Department's selection, commented:
Pavel's paper on quantile summary, co-authored with Professor Graham Cormode, was published in a top-tier conference on theoretical databases, Symposium on Principles of Database Systems (PODS) co-located with SIGMOD. The tight lower bound for quantile summaries proposed in this work is a deep theoretical exploration central to data management. The work led to a collaboration with Splunk, a US-based company that focuses on processing machine-generated big data. The follow-up paper has been accepted to the 2021 edition of the PODS conference. The result has a great potential for a broader impact. Josiah's work is on medical imaging. Together with Dr Sharon Collier and Professor Till Bretschneider, they proposed an enhanced 3D segmentation method (with a curvature-based enhancement term), which outperforms the best-of-breed random walker method in Dictyostelium image volumes. The work was published in a top journal, IEEE Transactions on Medical Imaging, with an impact factor of 6.68. Josiah presented his method at the Actin 2020 meeting and was awarded a prize for best imaging in a talk by the Royal Microscopical Society and has been featured in an article in their inFocus magazine.