Hardware-software Co-synthesis Research at Oxford

The work of the Hardware Compilation Research Group at Oxford is driven by the desire to close the gap between the study and practice of computing, and the study and practice of digital electronic engineering. Computing systems are becoming increasingly complex and often contain large amounts of both hardware and software. We believe that to manage eeectively the design of such systems mandates the use of increasingly comprehensive design tools which can close this gap. Errors in the design and understanding of systems usually congregate between the boundaries of subsystems , particularly when the subsystems have been engineered using diierent methods. This is certainly true when we consider systems with both hardware and software components because the subject of computing was split at a very early stage between the branches which deal with hardware and with software. At Oxford, we take the view that a single style of system description is needed, from which both hardware and software subsystems can be synthesised. We also believe that currently the only bodies of knowledge and practice which are capable of supporting the design of such very complex systems are mathematics and programming. It seems self-evident that the current styles of hardware description are completely inadequate to describe software. However, we believe that it actually is the case that the languages, tools, and practice of software engineering can be adapted to meet the challenge of hardware design. Indeed, there has been an obvious historical process over the last twenty years in hardware design which has resulted in the ability to describe higher and higher levels of abstraction in hardware description languages. These have tended to look more and more like programming languages as time goes by. The primaryàct of faith' of this research group has been to assume that at some point in the future we would all be routinely using programs as the single description of complete systems. Thus, we assume that many hardware-software systems will be created by programmers rather than mixed teams composed of programmers and electronic engineers. Since making this judgement, our work has been focussed on building the tools and methodologies to make this dream a practical reality. We believe that this stance makes our group rather diierent from some other groups who are using high-level descriptions to synthesise hardware. For the most part, we concentrate on the tools and methods necessary to allow programmers to build working …