To meet ever-increasing computing needs and overcome power density limitations, the computing industry halted simple processor frequency scaling and entered the era of parallelization, with tens to hundreds of computing cores integrated in a single processor, and hundreds to thousands of computing servers connected in a warehouse-scale data center. However, such highly parallel, general-purpose computing systems still face serious challenges in terms of performance, power, heat dissipation, space, and cost.

In this project we looked beyond parallelization to focus on domain-specific customization as the next disruptive technology to bring orders-of-magnitude improvement to important classes of applications. We focused our effort on revolutionizing the role of medical imaging by enabling techniques typically confined to research environments to be put to direct use for preventative, diagnostic, and therapeutic procedures. In later CDSC efforts, we expanded and extended the research to: 1) accelerator-centric architectures (ACAs) in which novel methodologies and algorithms to automatically extract accelerator building blocks (ABBs) in new application domains; and 2) compilation and runtime support for ACAs.

A focus of the CDSC during this phase were applications within healthcare. Key developments included an accelerated medical image processing pipeline; adaptive compressive reconstruction methods for CT and MR; and the “big data” challenges associated with next-generation sequencing analysis pipelines. Efforts were prototyped and validated on several types of programmable platforms, including state-of-art FPGAs and “supercomputer-in-a-rack” configurations.