> Well, I feel pretty dumb! I wrote a test program that spawns a couple of
> looping threads and sure enough it hammers both processes on NT using
> JDK1.3.
> So......
> My question now is, why, when I set-up a behavior that loops (attempting
> to reproduce the threading test above) does Java3D NOT use up all that
> extra processor time? Some people may be happy with that but I'm
> counting on Java and Java3D using everything it's got on every processor
> on every frame....
I'm not an expert on how Java3D does it, but there is a possible problem
with respect to OpenGL or any hardware accellerator. Basically, only
one thread may execute OpenGL commands at any one time, and its pretty
costly to switch the thread doing the writing.
Thus, unless you design things for multi-threading from the beginning,
it's difficult to get any gain from it. Notice that its quite possible
to do such designs - I belive the SGI Performer libraries does some pretty
advanced stuff in setting up pipelines to increase performance - but they
have to be an integral part of the design from the beginning.
I haven't looked at how the Java3D rendering pipeline looks, so I can't
say if this has anything to do with your problem... but it may help anyhow.
/Mats
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
