[ODE] Re: Different simulation results with different versions ofVisual C++?

John Donovan JohnD at MagentaSoftware.com
Mon Aug 8 14:12:50 MST 2005


> Have others found a way to work around this problem?  I need *exactly*
the same results from one simulation to another.

I haven't found a workaround, but I did notice the same problem when
running the same simulation twice at the same time. It all started
because I was gettinga diverging simulation with two slightly different
algorithms. So to see where it was diverging I created two worlds,
loaded the same data and rendered them simultainously. My problem turned
out to be a contact parameter issue and I fixed it quickly. But I then
noticed that running the two worlds _still_ diverged when run
simultainiously, but were identical when run individually. Because it
didn't really affect me I didn't bother following it up. But the only
conclusion I could come to was the random number generator was working
twice as hard when run in tandem, so generating slightly different
contact results.
So my advice would be to look at the RNG (there was some discussion on
it a while back on this list), and see if it is producing the same
results for each build.
Also, are you're running with a fixed timestep? With the different
optimisations from each compilier, you could well get a different
framerate every time.

-J

The content of this message and any attached file are confidential and/or privileged and are for the intended recipient only. If you are not the intended recipient, any unauthorised use, disclosure, copying, distribution or other dissemination is strictly prohibited. If you receive this message in error please notify the sender immediately by email, telephone or fax and then delete this email. Any attachment with this message should be checked for viruses before it is opened. Magenta Software Limited cannot be held responsible for any failure by the recipient to check for viruses before opening any attachment. Copyright in this email and attachments created by us belongs to Magenta Software Limited. Should you communicate with anyone at Magenta Software Limited by email you consent to us monitoring and reading any such correspondence. 






More information about the ODE mailing list