(The) stream supporting data flow system

Abstract

This thesis proposes a data flow system which supports the stream that has a more parallelism than any other data structure because it can execute operations in pipe-lined fashion. Compare to Weng``s abstract implementation, the proposed system is feasible for implementation while supporting the stream with the paged memory. With the 2-level hierarchical paged memory system, the stream can be processed without any restriction. The problem of excessive copying of data structures under the paged memory is solved by executing an instruction, which deals with a certain page, in the processor which already has that particular page. The simulation model for the proposed system is developed for GPSS simulation language and the simulation result showes that the proposed system is accurate.