Research
The project I am working on involves Coarse grained reconfigurable Architectures(CGRA). A CGRA is basically an array of processing elements such as arithmetic logic units(ALUs) or multipliers. The Processing Elements are connected with each other. CGRA’s are able to provide an impressive perforrmance improvements by exploiting parallelism in an application and are extremely well suited for applications that require very high throughput including multimedia signal processing, networking, and other scientific applications.
Coarse-grained Reconfigurable fabrics have great promise for achieving energy-efficient flexible designs for an application domain. In order to use these fabrics for practical applications, however, there is a great need to develop smart mapping algorithms to implement applications of interest onto these fabrics. This research is focused on discovering new mapping strategies for customized coarse-grained devices through crowdsourcing.
The main thrust of this research is to develop a science game to discover better mapping algorithms by making use of human intuition and ability to recognize patterns and opportunities even in complex problems. Players are presented with successively more difficult mapping problems in a game environment, and the vast dataset of players’ moves is analyzed to recognize common patterns used by successful game players. The insights gained from strategic moves humans make while solving problems based on their visual intuition and experience can be used to discover new mapping approaches that are beyond what can be conceived with traditional algorithms. New energy-efficient architectures and mapping algorithms that are developed in this research will spur development of a broad range of portable/wearable computing applications critical to health, safety and security, personal multimedia, and aerospace.
To CGRAs the compiler is an essential part in providing great performance, CGARs needs a compiler that is able to map applications on the a CGRA, and map using as little processing elements as possible. There are many compilers that are able to map applications onto a CGRA. What we are trying to do in this project is to compare traditional algorithims that map applications on CGRAs to algorithims generated by players through a free online web based puzzle game that challenges the player in solving graphs, these graph represents real applications that needs to be mapped onto an architecture. Players will try to shrink the their graphs as much as possible to obtain an higer score.