The design of nectar: a network backplane for heterogeneous multicomputers

Abstract

Nectar is a "network backplane" for use in heterogeneous multicomputers. The initial system consists of a star-shaped fiber-optic network with an aggregate bandwidth of 1.6 gigabits/second and a switching latency of 700 nanoseconds. The system can be scaled up by connecting hundreds of these networks together.The Nectar architecture provides a flexible way to handle heterogeneity and task-level parallelism. A wide variety of machines can be connected as Nectar nodes and the Nectar system software allows applications to communicate at a high level. Protocol processing is off-loaded to powerful communication processors so that nodes do not have to support a suite of network protocols.We have designed and built a prototype Nectar system that has been operational since November 1988. This paper presents the motivation and goals for Nectar and describes its hardware and software. The presentation emphasizes how the goals influenced the design decisions and led to the novel aspects of Nectar.

Perspectives on ATM switch architecture and the influence of traffic pattern assumptions on switch design

“…Interestingly, with only a few exceptions [YKS86], [F93], [ABC89] this class of switch is not well represented in the literature. Part of the rationale for the development of array switches was that not all of the crosspoints were needed [C53].…”

Communication in iWarp systems

“…Figure 2 shows a pathway between node (0,O) and node (1,3) that consists of four pathway segments: (0,O) -9 (OJ), (0,l) 3 (l,l), (1,l) + (1,2), and (1,2) 4 (1.3). The communication agents of nodes (OJ), (1,l) and (1,2) provide resources to implement the pathway segments that form this pathway, but these nodes cannot access any data that is transmitted over the pathway (nor can they use this pathway to send data). An iWarp network consists of nodes and pathways between these nodes.…”

Impact of protocol overheads on network throughput over high-speed interconnects: measurement, analysis, and improvement

“…Many high-performance computing systems (e.g., clusters [2] and grids [19]) employ such high-speed networks [1,3,23,42]. On the other hand, with the rapid increase of data volume to process, efficient data exchange between the computing nodes becomes more critical for many parallel/distributed applications.…”