3 ms·

A concession to the physical world?

• wizzwizz4 5y ago
  For networking, yes. For multicore, it's merely a concession to the fact that instructions on my architecture are variable-length, and there's a transparent cache mechanism (requiring knowledge of memory access patterns, which requires knowing the result of the computation ahead of time).
  • ProfHewitt 5y ago
    Because arbiters are used in communications among cores, a many-core computer has inherent indeterminacy.
    • wizzwizz4 5y ago
      Not necessarily? If you have a real-time OS and you write your program well, you can synchronise timings and have cores send messages to each other without queues. It's hard to write fast code that does that, but in narrow circumstances, it's possible (I'm thinking embedded applications) – and when it is possible, it's faster than the equivalent algorithm with indeterminacy. Indeterminacy slows things down. It's a concession.
      • ProfHewitt 5y ago
        Indeterminacy using many=cores speeds up processing. Enforced determinacy slows down processing.
        • wizzwizz4 5y ago
          And natural determinacy, if you can get it, speeds up processing more than indeterminacy. The useful question is “what's the lowest-level model I can usefully use?”, not “can I do it with indeterminacy?”. Do you think the above statement is wrong? If so, why?
          • ProfHewitt 5y ago
            Indeterminacy is crucial for next generation Intelligent System. See https://papers.ssrn.com/abstract=3581859 https://papers.ssrn.com/abstract=3581859