Speculative parallelism in BaLinda Lisp
C. K. Yuen, M.D. Feng, J.J. Yee · 2002
In a system supporting parallel execution, speculative processing can be used to increase the amount of parallelism. Speculative computations are started before it is known whether the results are needed. The system requires a method to favor the execution of more promising computations and to remove unwanted ones. Speculative processing can be equally or unequally prioritized. We describe the support for speculative parallelism in the language BaLinda Lisp, show its usefulness in applications to infinite data structure, heuristic search, and AND/OR parallel computation. We also discuss the condition and necessity for the early merging of the speculative tasks' environment.>