ResearchMany Cores Projects
OpenCL Performance Evaluation on Adapteva Epiphany Many-Core CPU
Evaluation of compiler and runtime performance of OpenCL benchmarks on 64-core Adapteva Ephiphany many-core CPU. Evaluation done with OpenCL benchmarks such as NBody algorithm, FFT, MatrixMultiply and CLSurf.
Low Power SuperComputing Cluster with ARM based nodes
Developed a Supercomputing facility comprised of 50+ nodes , each possessing a Dual Core ARM Cortex Series Processor. The cluster housed at MIT, Cambridge , MA is capable of achieving theoretical peak performance of 60GFLOPS for 50 Nodes. The power rating for 50 nodes being 225Watts. The Cluster can be powered by a Solar cell array. The Cluster is a potential candidate for a feature on the Green500 list.