Optimizing the heterogeneous network on-chip design in manycore architectures

Abstract

Current hybrid network-on-chip designs in manycore systems are agnostic to the application requirements and thus are provided for general cases. This results in high cost in the manycore systems design, wasted energy and performance. We observe that the cost of network-on-chip designs can be reduced by optimizing the application-specific traffic onto the system. This paper presents mincostflow-based heuristic algorithm (LINCA) that minimizes the quantification of hybrid routers corresponding to the application-specific traffic for manycore systems. LINCA guarantees the performance of hybrid networks on chip. Its results are validated against the manycore system architecture. Our evaluations show that LINCA can reduce significant cost of using hybrid routers in the manycore systems. It reduces cost by 84 percent on average across a variety of applications, compared with all of hybrid routers being deployed in the network without using optimization model.