This paper studies the memory coherence problem in designing said inaplementing a shared virtual memory on looselycoupled multiprocessors. Two classes of aIgoritb. ms for solving the problem are presented. A prototype shared virtual memory on an Apollo ring has been implemented based on these algorithms. Both theoretical and practical results show tkat the mentory coherence problem cast indeed be solved efficiently on a loosely-coupled multiprocessor.

Abstract not found

This paper describes a scheme for using cache-based hardware to provide simple and efficient message passing support for message-based software systems on a tightly-coupled, shared-bus multiprocessor. This approach is based on the utilization of the existing interprocessor communications medium, the shared bus, to effect the exchange of single-word messages. Communication between processes is accomplished over logical channels using simple, blocking send and receive primitives. The physical processor/channel interface is designed so that the message transfer primitives can be implemented as single machine instructions, namely store and fetch. Special-purpose caches, called message caches, mediate channel operations and effect the exchange of messages over the shared bus.

In this paper we analyze protocols for transmitting large amounts of data over a local are8 network. The data transfers analyzed in this paper are different from most other forms of large-scale data transfer protocols for three reasons: (1) The definition of the protocol requires the recipient to have sufficient buffers available ’ to receive the data before the transfer takes place; (2) We assume that the source and the destination machine are more or less matched in speed; (3) The protocol is implemented at the network interrupt level and therefore not slowed down by process scheduling delays. We consider three classes of protocols: stop-andwait, sliding window and blast protocols. We show that the expected time of blast and sliding window protocols is significantly lower than the expected time for the stopand-wait protocol, with blast outperforming sliding window by some small amount. Although the network error rate is sufficiently low for blast with full retransmission on error to be acceptable, the frequency of errors in the network interfaces makes it desirable to use a more sophisticated retransmission protocol. A go-back-n strategy is shown to be only marginally inferior to selective retransmission and is, given its simplicity, the retransmission strategy of choice. Our results are based on measurements collected on SUN workstations connected to a 10 megabit Ethernet network using 3-Corn interfaces. The derivation of the elapsed time in terms of the network packet error rate is based on the assumption of statistically independent errors. 1.