Abstract-We consider the problem of realizing several common communication structures in the all-optical Partitioned Optical Passive Stars (POPS) topology. We show that, often, the obvious or "natural" method of implementing a communication pattern in the POPS does not efficiently utilize its communication capabilities. We present techniques which distribute the communication load uniformly in the POPS for four of the most common communication patterns (all-to-all personalized, global reduction operations, ring, and torus). We prove that these techniques provide optimal performance in the sense that they minimize the time required to deliver the messages from each node to its neighbors.
This paper presents a partitioned optical passive star (POPS) interconnection topology and a control methodology that, together, provide the high throughput and low latency required for tightly coupled multiprocessor interconnection applications. The POPS topology has constant and symmetric optical coupler fanout and only one coupler between any two nodes of the network. Distributed control is based on the state sequence routing paradigm which multiplexes the network between a small set of control states and defines control operations to be transformations of those states. These networks have highly scalable characteristics for optical power budget, resource count, and message latency. Optical power is uniformly distributed and the size of the system is not directly limited by the power budget. Resource complexity grows as O ( n ) for tbe couplers, O ( n J n ) for transceivers, and O [ J tl log ( n ) ] for control. We present analysis and simulation studies which demonstrate the ability of a POPS network to support large scale parallel processing (1024 nodes) using current device and coupler technology.
This paper presents and analyzes a topological approach to providing multiple data channels using current I Introduction.The evolution of computer systems has, for the most part, been driven by an ever-increasing need for network throughput. Although the bandwidth wasted to control pure-electronic networks is not excessive, relative to network capabilities and system loads, these networks are incapable of efficiently acconiodating very large volumes of traffic due to several key physical link issues. These include driving power, rapacitive/inductive loading, and relatively low transmitter rates implied by sensitivity to noise.The second era in computer networking incorporated optical fiber link technology into existing multihop network designs [11, 1, 61. An order of magnitude increase in link throughput was avidable Further, the lack of reactive factors and a high noise immunity were well suited to the demands of the faster, larger, and more widespread environments. Such networks, however, still suffer from throughput bottlenecks and high latencies resulting from electronic/optical conversions and processing at intermediate hops. This paper studies a topological approach to providing multiple physical data channels. The Partitioned Optical Passive Stars (POPS) topology is an interconnection architecture that uses multiple nonhierarchical stars to achieve single-hop networks. It is an "all-optical" topology constructed exclusively with passive optical technology, and benefits from all the corresponding characteristics discussed, i.e., no intermediate electronic/optical conversions, no reactive factors and high noise immunity.A POPS topology is configured at design time to provide a fixed number of physically concurrent data channels, each of which is capable of high capacity in a circuit-switched system. The number of such channels is not absolutely limited, and is a key engineering tradeoff. This design flexibility provides for a customized optimization between lower total system complexity versus the combination of higher system throughput with both lower power budgets and lower network control overheads.After the introduction, the second section will discuss the components, parameters, and notation for a POPS topology. The third section will discuss the routing and scheduling of messages in a POPS topology. The fourth section will discuss the performance of POPS topologies in detail. The fifth section discusses issues related to scalability for very large system sizes. A conclusion summarizes key points about a POPS topology.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.