Abstract

The interconnection network comprises a significant portion of the cost of large parallel computers, both in economic terms and power consumption. Several previous proposals exploit large-radix routers to build scalable low-distance topologies with the aim of minimizing these costs. However, they fail to consider potential unbalance in the network utilization, which in some cases results in suboptimal designs. Based on an appropriate cost model, this paper advocates the use of networks based on incidence graphs of projective planes, broadly denoted as Projective Networks. Projective Networks rely on generalized Moore graphs with uniform link utilization and encompass several proposed direct (PN and demi-PN) and indirect (OFT) topologies under a common mathematical framework. Compared to other proposals with average distance between 2 and 3 hops, these networks provide very high scalability while preserving a balanced network utilization, resulting in low network costs.

Highlights

  • One current trend in research for the design of Exascale systems is to greatly increase the number of compute nodes

  • This paper shows examples of such topologies based on incidence graphs of projective planes and compares them with competitive alternatives

  • The target will be to maximize the number of terminals with minimum average distance and balanced link utilization, which is related to the generalized Moore bound

Read more

Summary

Introduction

One current trend in research for the design of Exascale systems is to greatly increase the number of compute nodes. The present paper deals with graphs attaining or approaching the generalized Moore bound [34]—which bounds the average distance for a given network size—while minimizing cost and power consumption. The target will be to maximize the number of terminals with minimum average distance and balanced link utilization, which is related to the generalized Moore bound. This allows to present a complete comparison, in terms of our power/cost model, of all these topologies, with special emphasis on the diameter 2 case.

Power and cost optimization
Network Dimensioning and Cost Model
Moore Bounds
Incidence Graph of Finite Projective
Projective Networks: A Topology
Modified Incidence Graph of Finite Projective Planes
Adverse Traffics
Topologies Near the Moore Bound
Topologies with Small Average Distance
Slim Fly
Projective Networks of Higher Average
Comparison of the Topologies
General Comparison
Projective Networks vs Slim Fly
Cases of Use
Indirect Networks
Findings
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call