tommyk@cs.glasgow.ac.uk (Tommy Kelly) (07/07/90)
It is perhaps obvious, but nonetheless true, that parallel machines owe their existence primarily to the fact that much of the work we demand of computers is based on algorithms which are to a greater or lesser extent, "parallelisable". Does anyone have any ideas, references, gut feel, etc for exactly how "greater" or "lesser" that extent is. How much parallelism is there in the tasks we run on computers? How does this vary with type of application? How much is it feasible to extract? How much is low-level fine-grain, and how much high-level? Any input would be appreciated - net or e-mail. tommy