Bsp Programming 1. the Bsp Model

The Bulk Synchronous Parallel (BSP) model provides a uni-ed framework for the design and programming of general purpose parallel computing systems. In this paper we describe some programming language developments which are currently being pursued as part of this new, uniied approach to scalable parallel computing. A parallel random access machine (PRAM) 4] consists of a collection of processors which compute synchronously in parallel and which communicate with a common global random access memory. A major issue in theoretical computer science since the late 1970s has been to determine the extent to which the idealised PRAM model can be eeciently implemented on physically realistic distributed memory architectures. A number of new routing and memory management techniques have been developed which show that eecient implementation is indeed possible in many cases 5, 10, 11]. The eecient implementation of a single address space on a distributed memory architecture requires an eecient method for the distributed routing of read and write requests, and of the replies to read requests, through the network of processors. Consider the problem of packet routing on a p-processor network. Let an h-relation denote a routing problem where each processor has at most h packets to send to various processors in the network, and where each processor is also due to receive at most h packets from other processors. Here, a packet is one word of information, such as e.g. a real number or an integer. Using two-phase randomised routing one can, for example, show that every (log p)-relation can be realised on a p processor hypercube in O(log p) steps. To deal with the problem of \hot spots",