A Flexible Multithreaded Programming Model and its scheduler for irregular workload

摘要

As we known, Chip Multiprocessors (CMPs) have become increasingly popular. However, it is difficult to program in parallelism with high performance on such platform, especially when we program with algorithms with irregular workload. Therefore, this paper presents a parallel programming model based on task pools, which provides a flexible framework for users to construct parallel program on multi-core processors. In this model, users only focus on data and task partition as programming, at the same time, the scheduler in the model handles the dynamic scheduling for irregular tasks. In this paper, the program interface and implementation are described with some examples to illustrate the model.