Question: RPC, threads, and concurrency - How are they related?
One minute response: Concurrency you should know about. Threads are a way to implement (or at least model) concurrency. RPC and concurrency are orthogonal topics, although the usual assumption is the RPC calls are synchronous (that is, non-concurrent); however, as we shall see, it is possible to define asynchronous (that is, concurrent) RPC calls.
Question: Why do we have a queue for timeouts in select() calls?
One minute response: I'm not sure to what this question is referring, but the usual reason for associating a timeout queue with select() calls is to implement a wake-up service for event-driven architectures. See tips 20 and 21 in Snader for details.
Question: Intel talks about their Hyper-Threading Technology. What is that? Hype?
One minute response: Hyper-threading (also called simultaneous multi-threading (SMT)) refers an architecture in which there are two or more CPUs (actually CPU cores) on the same chip. This architecture reduces the need to go off-chip when running co-operatively threaded applications (that is, concurrent programs), avoiding the off-chip performance hit; it's also a cheaper way to provide true concurrency than are multi-chip systems.
Whether or not hyper-threading is hype depends on your opinion of why symmetric multi-processing (SMP, not to be confused with SMT) failed. If you think SMP failed because of the off-chip performance hits incurred by co-operatively threaded programs or because of unfavorable cost-benefit ratios, then SMT probably isn't hype. If you think SMP failed because it's too hard to do, then SMT probably is hype.
Question: I imagine one could develop a version of probs that supported remote literal satisfaction? Or something like that.
One minute response: I don't understand the question.
This page last modified on 16 July 2003.