Abstract
This paper discusses a parallel graph reduction model and its implementation in the relation to the data flow computing scheme. First, a parallel graph reduction mechanism and its relation to data flow computing are discussed. In the discussion, the data flow computing model, which introduces by-reference concept is shown to be a natural implementation of parallel graph reduction model. Then, a practical implementation of a parallel reduction machine is presented. In the implementation, a cell token flow model is proposed. By using the cell token flow concept, a given data flow program can be transformed to an efficient multi-thread control flow program. Last, an overview of the machine architecture is described.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have