Hi Beppe, The best thing to do is to run a profiler on your code and see if there are any obvious bottlenecks. See ptII/doc/coding/performance.htm
You could also try the development version of Ptolemy II, available via http://chess.eecs.berkeley.edu/ptexternal Using ptexecute will usually be faster than using vergil because with ptexecute, no time is spent in the UI. Copernicus and the newer C codegenerator do not run under DE. I'm not sure I totally understand the issue you raise with simulation time being faster than real time. Make sure that synchronizeToRealTime is not set in the DE Director. DE is not particularly multithreaded. For a multithreaded model, you could try using PN, or the very experimental TimedPN director. _Christopher -------- Hi guys, I am working on a project on distributed sensors network and I am running a simulation under ptolemy. I use ptolemy under cygwin. The simulation (DE Director) involves Java classes that interface the actors with C code through JNI. it involves a variable number of "sensors" and actuating nodes. the problem is that I would loke to have the "simulation time" faster than (or equal to) the "Real Time" . For now, in 30 (RealTime)seconds I can simulate only (DE Dir) 8 sec. I noticed that: - (number of simulated seconds) is inverse proportional to (number of actors) - (number of simulated seconds) is inverse proportional to (number of messages) - the time spent in the code is really low. I run under a XEON biprocessor (each processor with 2 threads) and noticed that only one CPUTHREAD run at 90%, one at 10% and the other ones are unused. Are you aware od methods to run the simulation faster? I tried with no luck using ptexectute and running java simpleMoml (or something like that). Same results. I was also not able to run Copernicus, always some errors about libraries :( Maybe some compiling options to run pTolemy faster and use all the resources of the system. Or rearrange the code in order to better use the CPUs. Or...? If you want, I can made measurement and be more precise with numbers of messages, delays and so on.... Thanks a lot Beppe -------- ---------------------------------------------------------------------------- Posted to the ptolemy-hackers mailing list. Please send administrative mail for this list to: [EMAIL PROTECTED]
