SUMMARY A radio‐frequency pulse generator capable of inflicting variable but reproducible thermal injuries on tissue within the Sandison‐Clark rabbit car chamber is described. A technique facilitating the production of very small injuries at any chosen site within the car chamber is outlined. Some of the applications for the technique are discussed.

No abstract

The Melbourne University Optoelectronic Multicomputer Project is investigating dense optical interconnection networks capable of providing lowlatency data transfer of 32 or 64 bits. The networks we have developed do not need any optical switches and are therefore suited for implementation with state-of-the-art optical technology.Our research is currently concentrating on twodimensional topologies that broadcast data between the Processing Elements in each row and in each column. The simulated performance of random data transfer patterns indicates that multiple-broadcasting will be able to offer a cost-effective solution for low-latency interconnection networks in a massive parallel architecture. The question remains which implementation of multiple broadcasting will be the most successful.in this paper do not need message passing protocols, and can support efficient transfer of 32 or 64 bits of data.With an interconnection network capable of transferring small data items, it becomes possible for a parallel architecture to efficiently support small granularity parallelism when not enough large granularity parallelism is available in the application [12]. The minimum granularity of parallelism that can be supported depends on the data transfer delay of the network. It should be realised that to enable the exploitation of this small granularity parallelism, the network needs to support low latency data transfer even when all processors produce a data item at the same time [3]. This paper describes an example of a simple optical network capable of supporting the parallel transfer of data produced by every PE. We have simulated the performance of this network for different data transfer patterns. As will be explained, our simulator does not predict the transfer delay of a single data item, but instead evaluates the time needed to simultaneously transfer data items from all PE's to every PE that may need the data.