No abstract

In distributed systems uneven arrivals of the tasks may overload few hosts where as some of the hosts are lightly loaded. This load imbalance prevents distributed systems from delivering its performance to its capacity. Load balancing has been advocated as a means of improving performance and reliability of distributed systems. A new load balancing approach [14] is proposed to deal with this problem. In this paper we extend this model with multiple tokens. With some parameters in the algorithm set to intelligent values, the algorithm promises better load balancing results.

Abstract-Network-on-Chip (NoC) architectures have been widely adopted as the preferred solution to the communication challenges of System-on-Chip (SoC) design in the nanoscale regime. SoC designs often incorporate custom NoC architectures that do not conform to regular topologies. This requires the generation of a power and resource efficient interconnection architecture that can support the communication requirements for the SoC with the desired performance. Hence automated topology generation tools optimize for mutiple objectives like power and area to synthesize a NoC topology that meets these communication constraints. Clock-Domain-Crossings (CDCs) is an important consideration in fabric design that gets ignored in existing topology tools. CDCs add to latency and area, while also increasing verification and implementation effort. In this paper we propose a method to model and optimize the clock-domain-crossings (CDC) during NoC topology generation and prove that the underlying problem is NP-hard. We present and compare multiple approaches to model the CDC cost, including a novel fast heuristic. When applied within an existing topology generation tool our approach results in a 5% − 39% reduction in flop count of the resulting interconnect while still satisfying the original communication/performance constraints.

No abstract

No abstract