Assessing the benefits of fine-grain parallelism in dataflow programs
- Massachusetts Institute of Technology, Cambridge, MA (US)
A method for assessing the benefits of fine-grain parallelism in ''real'' programs is presented. The method is based on parallelism profiles and speed up curves derived by executing dataflow graphs on an interpreter under progressively more realistic assumptions about processor resources and communication costs. Even using traditional algorithms, the programs exhibit ample parallelism when parallelism is exposed at all levels. The bias introduced by the language Id and the compiler is examined. A method of estimating speedup through analysis of the ideal parallelism profile is developed, avoiding repeated execution of programs. It is shown that fine-grain parallelism can be used to mask large, unpredictable memory latency and synchronization waits in architectures employing dataflow instruction execution mechanisms. Finally, the effects of grouping portions of dataflow programs, and requiring that the operators in a group execute on a single processor, are explored.
- OSTI ID:
- 6524930
- Journal Information:
- Int. J. Supercomput. Appl.; (United States), Vol. 2:3
- Country of Publication:
- United States
- Language:
- English
Similar Records
Eps'88: Combining the best features of von Neumann and dataflow computing
Dynamic barrier architecture for multi-mode fine-grain parallelism using conventional processors