Abstract

Recent advances in system design provide increasing opportunities for rapid experimentation with task-oriented architectures, i.e., architectures designed to perform extremely well for a given application. Problems encountered in one such experiment indicate that, contrary to initial expectations, only a small fraction of the effort is required for hardware development. Most of the effort was devoted to algorithm analysis and restructuring, architecture design, systems programming, software and hardware debugging, and performance analysis. This paper presents a case study in task-oriented architecture, using the Harpy connected speech recognition system as the problem domain. This study highlights the limitations of current methods and presents a number of tools and techniques that are useful for experimentation with task-oriented architectures.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call