Prospects for optical interconnects in distributed, shared-memory organized MIMD architectures

The antipodes of the class of sequential computers, executing tasks with a single CPU, are the parallel computers containing large numbers of computin g nodes. In the shared-memory category, each node has direct access through a switching network to a memory bank, that can be composed of a single but large or multiple but medium sized memory configurations. Opposite to the first category are the distributed memory systems, where each node is given direct access to its own local memory section. Running a program in especi ally the latter category requires a mechanism that gives access to multiple address spaces, that is, one for each local memory. Transfer of data can o nly be done from one address space to another. Along with the two categorie s are the physically distributed, shared-memory systems, that allow the nod es to explore a single globally shared address space. All categories, the p erformances of which are subject to the way the computing nodes are linked, need either a direct or a switched interconnection network for inter-node communication purposes. Linking nodes and not taking into account the prere quisite of scalability in case of exploiting large numbers of them is not r ealistic, especially when the applied connection scheme must provide for fa st and flexible communications at a reasonable cost. Different network topo logies, varying from a single shared bus to a more complex elaboration of a fully connected scheme, and with them the corresponding intricate switchin g protocols have been extensively explored. A different vision is introduce d concerning future prospects of an optically coupled distributed, shared-m emory organized multiple-instruction, multiple-data system. In each cluster , an electrical crossbar looks after the interconnections between the nodes , the various memory modules and external I/O channels. The clusters itself are optically coupled through a free space oriented data distributing syst em. Analogies found in the design of the Convex SPP1000 substantiate the cl oseness to reality of such an architecture. Subsequently to the preceding i ntroduction also an idealized picture of the fundamental properties of an o ptically based, fully connected, distributed, (virtual) shared-memory archi tecture is outlined.