![]() A comparison is made between the MIT Tagged-Token Dataflow machine and the subject machine which presents a model for understandingthe cost of synchronization in a parallel environment. A prototypical architecture is described, and results of simulation studies are presented. The parallel machine language supports this notion via a large synchronization name space. In such an architecture,the units of scheduling, called scheduling quanta, are bound at compile time rather than at instruction set design time. These features are captured in the concept of a parallel machine language which can be grafted on top of an otherwise traditional von Neumann base. ![]() The analysis attempts to discover those featuresof the dataflow architecture, lacking in a von Neumann machine, which are essential for tolerating latency and synchronization costs. This paper examines the spectrum by proposing a new architecture which is a hybrid of dataflow and von Neumann organizations. At the opposite extreme are the von Neumann architectures which schedule on a task, or process,basis. Dataflow further offers a uniform synchronization paradigm, representing one end of a spectrum wherein the unit of scheduling is a single instruction. Iannncci HYBRID ARCHITECTURE IBM Corporation -andMIT Laboratory for Computer Science 545 Technology Square Cambridge, Massachusetts 02139 ABSTRACT Dataflow architectures offer the ability to trade program level parallelism in order to overcome machine level latencv. TOWARD A DATAFLOW / VON NEUMANN Robert A.
0 Comments
Leave a Reply. |