RASR

The Reconfigurable and Adaptive Systems Research (RASR) team develops enabling technologies for high performance deployable computing and reconfigurable supercomputing using Field Programmable Gate Arrays (FPGAs). We are interested in high density computation - systems, architectures, software tools, and applications mapped onto small form factor, deployable computing engines. We develop algorithms and prototype demonstration systems for real-time remote and autonomous processing of data gathered on land, in the air, or in space. Requirements for independent operation without human interaction and the necessity of sending critical information through low bandwidth channels, drive the project mission to find new solutions for real-time pattern recognition, detection, compression, and feature extraction, operating on a variety of time series, image, and video data types. Our research aims to provide a revolutionary leap (100X) in computing power for our data and compute intensive applications, enabling new science and providing new capabilties, and, futher, to obtain these performance gains within a framework that is inherently scalable, providing a long term solution to computing with exponentially more processors at exponentially smaller scales.
On-going projects include

• evolvable hardware for feature and object detection and tracking
• Reliability and fault management
• C-to-hardware compiler technology
• supercomputing application demonstrations on reconfigurable hardware
• system architecture studies of embedded acceleration engines
• adaptive techniques to manage available power for on-orbit processing
  

Team

Los Alamos Collaborators

External Collaborators

Our team at Los Alamos also has on going collaborations with Xilinx, Brigham Young University, University of Rennes, Virginia Polytechnic Institute and State University (Virginia Tech), Cell Matrix Corporation, Information Sciences Institute (East) of University of Southern California, University of California at Riverside, and others.