This thesis investigates an energy-efficient, high-performance, and high-density reconfigurable computing system for cloud data centers. By introducing decentralized fabric management agents for FPGAs and network topology configuration of software-defined compute platforms, we demonstrated flexible resource allocation at a significant reduction of the number of CPUs required. Compared to existing solutions, the proposed system also improves the data-path and control-path performance of distributed applications, as well as the infrastructure compute density.
«
This thesis investigates an energy-efficient, high-performance, and high-density reconfigurable computing system for cloud data centers. By introducing decentralized fabric management agents for FPGAs and network topology configuration of software-defined compute platforms, we demonstrated flexible resource allocation at a significant reduction of the number of CPUs required. Compared to existing solutions, the proposed system also improves the data-path and control-path performance of distribut...
»