Matthias Gries, Kurt Keutzer (editors), Springer Verlag, June 2005, ISBN: 0-387-26057-9
When I was at AT&T Bell Laboratories, an integrated circuit designer, Mark Vancura, once told me: "A good talk is one where I go back to my office and do something differently." Over time this has also become my touchstone for quality research in electronic design automation and integrated circuit design. Application specific instruction-set processors (ASIPs) are as old as computing itself. Techniques and tools for the design of ASIPs have been an active research area for fifteen years now. The dream of the automatic generation of hardware and software -- including simulators, assemblers, and compilers -- has been commercially realized. The utility of these tools in enabling the exploration of the architectural design space has been demonstrated. More ASIPs are designed every year and ASIPs have become the building block of the highest performance programmable platforms of our era. For all this, the design of ASIPs is typically a chaotic process, and the commercial success of the result of this process is unpredictable. Working with our colleagues at CoWare and Tensilica we have developed a methodology that we hope will bring some rigor to ASIP design. In short, if you're working with ASIPs take this book home and read it. We hope that when you head back to your office you'll end up doing something differently.
Back to Mescal book overview