The dream of the computer systems designers always was to transform a design specification to a running system automatically, without (or with minimal) human/designer intervention. New technological developments that bring together multi-core computing with FPGAs create huge opportunities. The most recent acquisition of Altera by Intel Corporation indicates that some key pre-conditions for achievement of this dream are created. However, software technologies and their link with custom computing machines that can be implemented in FPGAs are well behind the needs of future applications and systems. Examples of application domains for such systems are deep learning, big data and real-time systems. The project will look at how complex computations can be specified on high level and transformed into running system on a mixed platform that includes heterogeneous multi-core capabilities with FPGAs. A Network on Chip (NoC) will be used as the basic infrastructure that enables connecting heterogeneous computing resources of the platform. Depending on students capabilities there are multiple potential research directions.
Undergraduate
NoC-based platform that combines standard and custom processors and application functions implemented in hardware. Basic compilation technology that allows the use of the platform in a selected target application. Depending on the team members project can be customised to go in a specific research direction.
At least one student has to have experience with FPGAs and digital hardware design. Prerequisite can be COMPSYS305, co-requisite COMPSYS701. Also, good command of Java is preferable.
Lab allocations have not been finalised