We have measured the gap between FPGAs and ASICs in a fairly rigorous way, in order to understand the forces and opportunities for closing that gap.
We are also interested in automating the creation of FPGAs themselves, by developing tools that can automate the both the electrical and physical design of an FPGA from a relatively simple, textual description of an FPGA architecture. I believe this will work will give us insight into FPGA architectures.
A third area of interest is to build flexible hardware systems, and then use them for novel applications that can benefit significantly from that flexibility, in areas such as computer vision, graphics, biomedical and bioinformatics.
Another way to improve FPGAs is to build software specific to application areas (such as Processors, for example) that constructs the "perfect" hardware for a given application, a key capability of FPGAs.
Below is a listing of the various sub-disciplines of my research on
FPGAs, and links to other pages describing these topics further.
|CAD for FPGAs|
|Field-Programmable Systems and Applications|
|Soft Scalar Processor Architecture and Downloadable Code for Soft Processors|
|Soft Vector Processor Architecture and Downloadable Code for Vector Processors|
|Automated FPGA Creation (Electrical Design and Layout) from an Architectural Specification|
|Synthetic Benchmark Generation and Circuit Characterization|