NSF Award
1046946
This Small Business Innovation Research Phase I project aims to prove the technical and commercial feasibility of a virtualization technology for multicore computer systems, from mobile clients to server clouds. Although multicore processors offer tremendous performance potential, most existing applications are sequential and many new applications are written in sequential languages because the majority of mainstream programmers are most comfortable with sequential programming. Even when applications are parallel, achieving high performance is tedious as programmers must customize their code to each target system. As the number of cores per processor continues to increase, the key challenge is how to make use of these cores to deliver increased performance. Multicore processors execute code in parallel, but writing and debugging explicitly parallel programs is significantly more difficult than writing and debugging a sequential program. The goal of this proposal is to demonstrate that the proposed technology brings realizable, scalable performance to multicore systems which will leverage the current predominant architecture and bring about enhanced performance promised by multicore platforms.<br/><br/>The superior performance and energy efficiency delivered by the proposed technology will, if successfully deployed, enable innovative applications and services that were not possible on a single-core platform. The technology allows software developers to continue to focus on such applications and services rather than dividing their efforts with the extraction and debugging of parallelism for the underlying hardware. Software users can readily tap into expertise in program optimization to accelerate high-value applications. The technology provides manufacturers of mobile devices and PCs with the ability to dramatically improve the end user experience, to differentiate their products in these fiercely-competitive markets.
