Distributing data packets according to processing bandwidth

摘要

Data packets of an incoming stream are distributed in whole or in part to processing elements in dependence on the data processing bandwidth of the elements. In a scalable data processing architecture described (Fig 6) the elements are identical and are arranged in a cluster to provide, multi-threaded, parallel processing. They all execute a single instruction stream, have their own embedded memories, are able to make independent data-driven decisions and may be assisted by specialised, pipelined, hardware engines for table lookup and other functions.