Seeking Solutions in Configurable Computing

38 Computer Seeking Solutions in Configurable Computing C onfigurable computing systems combine programmable hardware with programmable processors to capitalize on the strengths of hardware and software. Often these systems must also address the difficulties of both hardware and software, because they mix the technology. While the origins of configurable computing go back at least 30 years, the past eight years have brought about a significant increase in research activity. Since at least 1989, 1 configurable computing systems 2 have demonstrated the potential for achieving high performance for a range of applications, including image filtering, convolution, morphology, feature extraction, and object tracking. Researchers have developed prototype systems that achieve performance an order of magnitude higher than more conventional approaches for a number of applications. However, realizing this potential outside of the laboratory has proven difficult because these systems rely on manipulating low-level abstractions—digital circuits, for example—and thus require highly skilled developers. The earliest configurable computing machine was likely proposed, designed, and implemented by Gerald Estrin at UCLA in the early 1960s. 3 Estrin proposed the " fixed plus variable structure computer, " which dedicated hardware to both an (inflexible) abstraction of a programmable processor and a (flexible) component that implemented digital logic. This basic architecture, which supports programmed hardware and software, is at the core of all subsequent configurable computing systems. Unfortunately, Estrin's architectural concepts were well ahead of the enabling technology, and he was only able to prototype a crude approximation of his vision. Many of the concepts that are now being discovered by the configurable computing community lie quietly unheeded in Estrin's early publications. The enabling technology behind the renewed interest in configurable computing is the availability of high-density VLSI devices that use programmable switches to implement flexible hardware architectures. These chips contain memory cells that hold both configuration information for the programmable switches and state information for active computations. Before programming, the chips present a partial architecture, which is then refined according to the configuration information. The configured device provides an execution environment for a specific application. The most common devices used for configurable computing are field programmable gate arrays. FPGAs present the abstraction of gate arrays, allowing developers to manipulate flip-flops, small amounts of memory , and logic gates. Figure 1 illustrates the basic architectural components of all configurable computers. This highly abstracted model allows a wide range of design choices, all of which revolve around three …