Center for Hybrid Multicore Productivity Research (#1031999, #1032040, and #1032077)
This proposal seeks funding for the Center for Hybrid Multicore Productivity Research located at the University of Maryland Baltimore County (UMBC as the lead institution) with research partners at the University of California, San Diego (UCSD) and Georgia Tech (GT). Funding Requests for Fundamental Research are authorized by an NSF approved solicitation, NSF 10-507. The solicitation invites I/UCRCs to submit proposals for support of industry-defined fundamental research.
The proposal prototypes a new virtual environment, where based on new methods heterogeneous computing resources are integrated in the virtual world to mimic dynamic situations. The new data and dynamic situations are continuously evolving based on the availability of information from various sources, and thus updating the virtual world representation, which can be about disaster situation or large event of societal and environmental importance. The proposal combines methods to use cloud services, virtual world servers, compute accelerators and client systems ranging from multi-core desktops to mobile phones to build the virtual world test bed. The PIs are very well qualified to conduct the proposed research. The proposed activities have sufficient access to the necessary resources.
This project will allow for synthesis of best current practices in divergent areas of multicore computing, data assimilation models and knowledge representation while exposing areas requiring new approaches to better utilize emerging computing paradigms to solve urgent broad social needs. The work will be readily applicable to develop future event response situations, including natural or manmade disasters, e.g. recent Haiti earthquake. All three universities have active programs for female, international and African-American students and cooperative programs with local schools to encourage and mentor high school and undergraduates from underrepresented groups to go on to university and graduate school in science or engineering, including computer science. Results will be disseminated by reports and meetings. The developed software will be made available under the same public license as Google Android.
The goals of this project were to use mobile devices to in order to assess damage and provide help to people affected by human-made or natural disasters. We did this for three different scenarios: the Haitian earthquake of January 12, 2010; the Deepwater Horizon oil spill of April 20, 2010; and the 2011 T?hoku earthquake of March 11, 2011. For each of these we developed approaches to search through social media databases such as flikr and twitter for metadata tags which would either mention these sites or have geo-location information. Finding this information, we would locate the social media file into a geographic specific location, correlated to the GIS information in a virtual world simulation. Additionally, with the Deepwater Horizon oil spill we developed a simulation in the virtual world utilizing the GNOME predictive oil spill model from NOAA. From this we noted that there were boundary discrepancies between the GIS models we were using in our virtual world simulation and the lands/water boundary utilized in GNOME. We were able to see good correlation between the GNOME simulation and observational data uploaded to social media websites. Georgia Tech served the role of being the computer architects/hardware experts in this project. We have first characterized the computation required, and then using the Texas Instruments OMAP4—a very popular mobile processor used in many of today's smartphones—we have characterized the computational requirements for running advanced pieces of the algorithms on cellphones. This has resulted in some interesting observations, such as the IVA processor on the OMAP, which was originally designated in the chip as being special-purpose for communications, is a better computational platform than the two-processor ARM Cortex-A9 cluster. However, the ARM Cortex-A9 cluster has media excelleration support via the NEON application extensions. With these, the ARM processors are actually more power efficient for computation. Intellectual merit: We developed new methods and procedures for using data and mobile phones to help in disaster relief. Broader impact: It is not an understatement that this work will have a broad and profound impact on reducing the suffering of those impacted by natural or human-made disasters.
