Output details

Automatic Extraction of Coarse-Grained Data-Flow Threads from Imperative Programs

<08> In a context where sequential performance stagnates and parallel execution platforms become increasingly available, this paper focuses on a new technique for automatically extracting data flow parallelism from sequential programs. This paper is significant because it represents a first step towards automatic parallelization of (unrestricted) imperative programs. Furthermore, data flow parallelism is more easily exploited on current architectures than other forms of parallelism because it puts less pressure on the available memory bandwidth.