Dynamic Dataflow Architecture Research Articles

Double Input Operators of the DF KPI System

Dataflow architectures can be used advantageously for computation-oriented applications that exhibit a fine grain parallelism. The implementation of the dataflow computer architecture depends on the form of execution of the dataflow program instructions, which is implemented as a process of receiving, processing and transmission of data tokens. The architecture described in this paper belongs to a class of dynamic dataflow architectures with direct operand matching. The concept of direct operand matching represents the elimination of the costly process (in terms of computing time) related to associative searching of the operands. This process is associated with the processing units of the proposed system. The processing units are designed as a dynamic multifunction pipelined unit of five segments, Load-Fetch-Operate-Matching-Copy. This pipeline stages handle processing of operand matching of dataflow operators. From the many types of operators, this paper describes microprogram managing for double input operators.

Distributed event-driven simulation of a novel dynamic dataflow multiprocessor system—PATTSY

This paper concerns the implementation of a distributed eventdriven simulator, which elicits some of the salient features of a novel dynamic dataflow architecture, PATTSY. The simulator has been developed on a network of Sun workstations operating under the Unix environment. The results from the simulation of PATTSY which are presented deal with the following aspects of the system: processor configuration, maximum buffer sizes, number of subsystem nodes, overall execution time, and locality effect. We have demonstrated that to achieve the best execution time for a dataflow program, the number of nodes within the system and the type of mapping scheme used are particularly important. In addition, we observe that (1) a large number of subsystem nodes will allow more actors to be fired concurrently, but the communication overhead in passing data tokens to their destination nodes causes the overall execution time to increase substantially, and (2) the mapping scheme employed has to exhibit a strong locality effect in order to allow efficient execution. The relationship between the mapping scheme employed and locality effect in a program are discussed.

A data driven hybrid computer architecture

The combination of dataflow and von Neumann execution models is a recent trend in designing high speed computers. In this paper, a data-driven hybrid computer architecture is presented. Dynamic data-driven execution principle, instead of program counter, is used to control the execution of instructions in a von Neumann style pipelined architecture. Unlike normal dynamic dataflow architecture, data are explicitly stored in memory and their memory locations are used as tags. Matching operation is accomplished by a simple comparison of two counters and no special matching unit is required. With an ideal memory system, no bubble may occur in the pipe if sufficient parallelism exists in the program. Furthermore, multiple memory modules and short-circuit scheme are used to fulfill simultaneous memory requests. An extensive simulator has been designed to evaluate the proposed architecture. The experimental results show that the proposed architecture is promising.

The program dependence web: a representation supporting control-, data-, and demand-driven interpretation of imperative languages

The Program Dependence Web (PDW) is a program representation that can be directly interpreted using control-, data-, or demand-driven models of execution. A PDW combines a single-assignment version of the program with explicit operators that manage the flow of data values. The PDW can be viewed as an augmented Program Dependence Graph . Translation to the PDW representation provides the basis for projects to compile Fortran onto dynamic dataflow architectures and simulators. A second application of the PDW is the construction of various compositional semantics for program dependence graphs.

Analysis of computation-communication issues in dynamic dataflow architectures

This paper presents analytical results of computation-communication issues in dynamic dataflow architectures. The study is based on a generalized architecture which encompasses all the features of the proposed dynamic dataflow architectures. Based on the idea of characterizing dataflow graphs by their average parallelism, a queueing network model of the architecture is developed. Since the queueing network violates properties required for product from solution, a few approximations have been used. These approximations yield a multi-chain closed queueing network in which the population of each chain is related to the average parallelism of the dataflow graph executed in the architecture. Based on the model, we are able to study the effect on the performance of the system due to factors such as scalability, coarse grain vs. fine grain parallelism, degree of decentralized scheduling of dataflow instructions, and locality.

A tagged token dataflow machine for computing small, iterative algorithms

A "buildable" dynamic tagged token data flow architecture tailored for small, highly iterative algorithms such as the Mandelbrot set is presented. Historically, dataflow machines have required a major hardware effort to construct. As a result, very few real machines have been built. In contrast to these large scale machines, this project was intended to be built at minimal expense and with "off-the-shelf" parts. A novel insight into the process of executing a dataflow graph is presented. This insight has resulted in a new architecture utilizing a "pseudo-program counter". This scheme allows a significant reduction in the width of the data paths throughout the PE interconnection network. To further reduce the required hardware, the traditional associative matching store has been eliminated in favor of a more brute-force approach that is workable for the class of algorithms considered for execution on this machine.