Grant: Research Tools and Instruments - Category 1 (<$150,000)

This proposal seeks funding for a cluster to support two complementary research programs. The first research project studies Grid Computing. Current Grids use software to enable controlled resource sharing, facilitating the creation of "virtual organizations" (VOs). Despite remarkable progress, significant human and hardware resources are currently allocated to support each VO. The large number of functionally independent services that must be deployed, configured, and reliably maintained is duplicated in each VO. To reduce this high deployment and operational burden, and reap specialization and economies of scale benefits, Ripeanu's research aims to enable VOs to outsource common services to utility-like service providers. More specifically, we design a set of scalable utility service implementations and provide the hosting infrastructure required to deploy and operate these services.       The second research project studies computer architecture. As semiconductor scaling increases the number of transistors that can be integrated on a single chip, communication latencies and bandwidths scale at different rates in relation to the available computational resources. Recently these changes have lead to increasingly complex and poorly understood system level behavior in high performance microprocessors and single chip multiprocessors. The core objective of Aamodt's research program is to develop and validate modeling techniques to optimize the microarchitecture design of such processors and make eliminate the bottleneck this stage of development represents for innovation in computer system design.       While these programs are largely unrelated in terms of research objectives, they make a compelling case for a shared infrastructure: they guarantee high utilization of the acquired resources and inform each other's design decisions.