The authors have designed, simulated, tested, and implemented a customized RISC processor to facilitate associative search and aggregation operations for relational database systems. They show that rapid design of application-specific integrated circuit processors is feasible with the right computer-aided design tools and design methodology. The rapid prototyping of the data filter processor consists of three phases: exploration of architecture alternatives, architecture design simulation, and implementation. Prolog is used as the hardware prototyping language. The fabricated chip has 91,000 transistors, 171 pins, and a cycle time of 54 ns.
The authors have performed neither a sizing sensitivity analysis nor a timing sensitivity analysis, so it is not known how to improve the chip effectiveness by such means as increasing the transistor density on the chip or reducing the chip size, cycle time, or power requirements. Indeed, at sufficiently high speeds, software and hardware metastability will each have several adverse effects on chip performance. All in all, I highly recommend the paper to anyone interested in the fast design and implementation of small, high-speed database systems capable of handling thousands of complex search requests concurrently.