PGR Seminar – Sing Hin To and Tianchen Wu
You are warmly invited to the next PGR seminar:-
Tuesday 19/05/2026
15:00-16:00
JC 1.33A
Speaker 1: Sing Hin To
Title: HABITAT – Hardware-accelerated binary translator
Abstract: Emulators have been essential in supporting legacy systems, future prototypes, and programs for other architectures. Examples include BlueStacks for Android, PCSX2 for PS2, and Rosetta on Apple machines. However, software-based solutions suffer from performance and scalability issues, which implies the need for powerful and expensive systems. This project seeks to enhance the performance and accessibility of emulators by adding a dedicated chip to accelerate certain stages of emulation.
Bio: Jason is a first-year PhD student supervised by Dr Tom Spink. His research investigates the potential improvement in emulation technology & research in instruction set architecture (ISA) with specialised hardware. Jason is currently interested in system programming, OS, and the acceleration /optimisation of programs.
Speaker 2: Tianchen Wu
Title: Towards Automated Generation of Benchmark Instances with Diverse Solver Performance
Abstract: As with most domains, high-quality benchmarking data are a vital resource in the domains of combinatorial problem solving and operations research. However, hand-crafting example problems usually require domain specific knowledge as well as offer limited coverage over the problem space. Consequently, automated benchmark data generation has been widely studied across disciplines. We study AutoIG in particular, a constraint-based automated instance generation framework, which has the key limitation that it does not explicitly encourage diversity among the generated instances. We introduce and investigate some simple mechanisms for promoting diversity in the performance space and provide a small case-study using two problem classes and two solvers to discuss some of the open challenges.
Bio: Tianchen Wu is a first year PhD student supervised by Dr Nguyen Dang and Prof Ian Miguel. His research focuses on automating the generation of benchmark data for combinatorial problems, with an emphasis on domain-agnostic methods for creating diverse datasets.