Cloud/High-Performance Computing (HPC) Architectures by Design

a. Cloud/High-Performance Computing (HPC) Architectures by Design Innovation in cost-effective cloud/high-performance computing (HPC) architectures by design is needed to enhance traditional capabilities in remote sensing and signal processing capabilities and enable non-traditional methods (e.g., machine learning) for feature engineering and detection. There is a growing opportunity for expanded mission capability through the utilization of increasingly-available public and commercial imagery and sensor data, as well as open-source text streams, including social media, newsfeeds, scientific literature, and public records. There is also a growing number of HPC-based system models (deterministic and optimization-based) that need to be cost-effectively integrated with data analytics through cost- and time-optimized cloud/HPC architectures. Successful proposals will develop a system concept and software architecture for a proposed cloud and HPC combined architecture that optimizes cost-to-solution and time-to-solution, and should include at least two diverse data streams and one HPC-based systems model for demonstration of a limited-functionality prototype. Phase I deliverable is a final report that describes system concept, software architecture, algorithms, and results from limited functionality prototype demonstration. Questions Contact: Angie Sheffield, angela.sheffield@nnsa.doe.gov