TowerTest

Hi all!

Im not sure if you have perviously noticed, but the tower test was a bit…umm…crap! Because if you add more blocks to the tower, odejava would start crashing due to it not having enough stack size.



Ive modified the test to create a new thread and start jme in that thread instead of the main thread, and now, we have:



78 blocks

1 pyramid

1 sphere (heavy)

9 balls



total of 89 dynamic objects all colliding with each other in the same scene running at 160 FPS on my crap machine! :smiley: And when the ball breaks the pile up (i.e. less collisions are happening), it goes back to 260 ( a maximum of 300 without physics), so the physics with 89 dynamic objects only uses 40 FPS! Wooo! :smiley:



Im gonna go make me a stress test!



DP

I get a vm crash, it start and the vm fails completely I had similar issue with the particle editor jme doesn’t seem to handle handle threads consistentlyl

Humm, works well here (between 300 and 400 fps)…

@mcbeth: did u try -Xss1M as specified by the javadocs of those tests?



The ode native stack needs more space for more objects…



DP

"DarkProphet" wrote:
@mcbeth: did u try -Xss1M as specified by the javadocs of those tests?

The ode native stack needs more space for more objects...

DP

sorry about that, didn't read them, will get back to you later if still have issues