The Data-Driven Multithreading Programming Model Based on DDF

Currently the compiler technology is far from the need of fully exploring the benefits of chip multi-core architecture of nowadays microprocessor. To further exploit parallelism inside real-world applications, explicit expression of parallelism at thread level will be an important approach. Based on dynamic data flow (DDF) model, this paper presents a multithreading programming model. Using the model parallel execution, analyzability, testability of a multithreading program can be guaranteed by the property of the model of computation. With the programming model, a task can be expressed by multiple code blocks allowing execution in parallel and on different processor architectures, thus such a technique can also be applied in pervasive computing area in future.