Wormhole routing techniques for directly connected multicomputer systems

  • Authors:

  • Prasant Mohapatra

  • Affiliations:

  • Iowa State Univ., Ames, IA

  • Venue:
  • ACM Computing Surveys (CSUR)
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

Wormhole routing has emerged as the most widely used switching technique in massively parallel computers. We present a detailed survey of various techniques for enhancing the performance and reliability of wormhole-routing schemes in directly connected networks. We start with an overview of the direct network topologies and a comparison of various switching techniques. Next, the characteristics of the wormhole routing mechanism are described in detail along with the theory behind deadlock-free routing. The performance of routing algorithms depends on the selection of the path between the source and the destination, the network traffic, and the router design. The routing algorithms are implemented in the router chips. We outline the router characteristics and describe the functionality of various elements of the router. Depending on the usage of paths between the source and the destination, routing algorithms are classified as deterministic, fully adaptive, and partially adaptive. We discuss several representative algorithms for all these categories. The algorithms within each category vary in terms of resource requirements and performance under various traffic conditions. The main difference among various adaptive routing schemes is the technique used to avoid deadlocks. We also discuss a few algorithms based on deadlock recovery techniques. Along with performance, fault tolerance is essential for message routing in multicomputers, and we thus discuss several fault-tolerant wormhole routing algorithms along with their fault-handling capabilities. These routing schemes enable a message to reach its destination even in the presence of faults in the network. The implementation details of wormhole routing algorithms in contemporary commercial systems are also discussed. We conclude by itemizing several future directions and open issues.