Network-based multicomputers

Abstract

Multicomputersbuiltaround a general network are now a viable alternative to multicomputersbased on a system-speci~c interconnect because of architectural improvements in two areas. First, the host-network interface overhead can be minimized by reducing copy operations and host interrupts. Second, the network can provide high bandwidth and low latency by using high-speed crossbar switches and efficient protocol implementations.While still enjoying thejexibility of general networks, the resulting network-based multicomputers achieve high performance for typical multicomputer applications that use system-specijic interconnects. We have developed a network-based mtdticomputer called Nectar that supports these claims.This paper argues that it is feasible to build highperformance network-based multicomputers that use general networks instead of system-specific interconnects. Such a multicomputer is able to use em"sting hosts, including workstations and special-purpose processors, as its processors. While enjoying a high degree of flexibility, the underlying network can have performance comparable to that of a dedicated interconnect. This is true both for large messages,where bandwidth is important, and for small messages, where software overhead is typically the limiting factor. As a result, the types of applications that run on existing mult.icomputers also run efficiently on network-based multicomputers. At the same time, network-based multicomputers are able to take advantage of rapid advances in network and processor technology.The Nectar project is one of the fist attempts to build a high-performance network-based multicomputer. Nectar is composed of a high-bandwidth crosspoint network and dedicated network coproeessors. A prototype system using 100 Mb/s (megabits per seeond) links has been operational since early 1989. The system has currently 26 hosts, including a connection to a CRAY Y-MP via a 26 kilometer singlemode fiber link. The Nectar prototype has been used as a vehicle to study architectural issues in network-based multicomputers and high-speed networks. This paper is based on insights and experiences gained from the Nectar prototype.To further pursue our ideas, we are collaborating with an industrial partner (Network Systems Corporation) to develop a gigabit Nectar system capable of sustaining gigabit per second end-to-end communication.This new system will support the 800 Mb/s HIPPI (High-Performance Parallel Interface) ANSI standard, and will also have a SONET/ATM 664

Analyzing communication latency using the Nectar communication processor

“…Multicomputers that use existing hosts and a general network are an attractive architecture for many applications [11], and they are one of the main motivations for improving the performance of networks [1]. The transport protocol does the necessary processing, hands the packet off to the datalink protocol, which places it on the wire, When the message arrives on the receiving CAB, the SPARC is interrupted, and the datalink protocol places it in a mailbox using Begin_Put.…”

Analyzing communication latency using the Nectar communication processor

“…This overhead however is, or should be, of the order of 10s of microseconds [7], and does not account for the order of magnitude difference in latency. In this paper we identify other sources of overhead based on measurements collected on Nectar.The Nectar network [3,11] consists of a high-bandwidth crosspoint network (Figure 1-1). Host are connected to the network through Communication Acceleration Boards (CABS) that are responsible for protocol processing.…”

“…The Nectar network [3,11] consists of a high-bandwidth crosspoint network (Figure 1-1). Host are connected to the network through Communication Acceleration Boards (CABS) that are responsible for protocol processing.…”