We have a Java-embedded ECLiPSe 6.1 process running on Win7 Pro 64-bit 2.67GHz that takes more than two weeks to run, but all the while the CPU and RAM utilization is quite low. We are looking for ways of increasing CPU utilization in order to shorten the run time. One of our colleagues speculated that "you are doing lots of read or write to your database on the disk and that is causing the IO bottleneck." Is any of the ECLiPSe database kept on disk instead of RAM? Any suggestions on where our bottleneck might be? Regards, David btw, my own speculation is that, because we have 250,000+ facts with the same functor, that the long runtime is due to serially attempting matches across most or all of those facts. We are looking into storing few facts or using more functors, but I've wondered if Google's parallelization of graph search algorithms in BigQuery might someday come to ECLiPSe to allow parallel search for backchaining.Received on Tue Sep 04 2012 - 02:53:15 CEST
This archive was generated by hypermail 2.3.0 : Tue Apr 16 2024 - 09:13:20 CEST