Inside Sabertooth
Learn how Sabertooth uses 3ds Max to create 3D interactive projects, including HBO Go’s Game of Thrones interactive experience
  • 1/3
You are here: Forum Home / Autodesk® Softimage® / ICE - Interactive Creative Environment / ICE memory leak w instance group?
  RSS 2.0 ATOM  

ICE memory leak w instance group?
Rate this thread
 
28897
 
Permlink of this thread  
avatar
  • Total Posts: 22
  • Joined: 20 February 2008 08:04 PM

i have an ice point cloud and an instance group plugged in.
now when i render, softimage keeps filling up my memory and and at some point just crashes.
is there anyone else with the same problem?
help is extremely welcome

thanks guys
r



Replies: 0
avatar
  • alanf
  • Posted: 03 December 2008 04:07 AM

Are you using an “Instance Shape” node (plugged to a self.shape setdata) or a “Set Instance Geometry” compound?

If you’re using one, try the other. The Instance Shape node should be lighter, but Set Instance Geometry uses (internally) “mentalray assemblies”, which get written to disk and then loaded at rendertime. (Maybe that manages memory better? I’m not sure.)



Replies: 0
avatar

hi alanf
thanks. instance shape is the lighter one. my question now is why do you even need the instance geometry and even with the instance shape xsi keeps loading into the memory and at some point keeps crashing or just wont do anything.

thanks for your fast response
r



Replies: 0
avatar

I would like to hear an answer to this as well. I sent a scene off to Softimage and they said they couldn’t reproduce the leak. Not sure how that’s possible, but the memory footprint just grows and grows until the machine becomes unresponsive.



Replies: 0
avatar
  • T.I.M.
  • Posted: 04 December 2008 08:21 AM

We had the exact same problem with one project. So we had to use batch rendering and make the render to restart in every 10 frames to clear the memory even that didn’t work very well. It was nightmare. We sent the scene to Softimage support. I hope they take this problem seriously and fix it soon.



Replies: 0
avatar

Yes. Exactly the same for us. Limiting our packets to 10 or so frames and memory didn’t always release as expected. In the end I had to babysit several 1500 + render sequences. It’s good to know others have seen this too. Let’s keep our fingers crossed on a fix.



Replies: 0
avatar

alright
good to hear i am not getting insane....;)
i hope xsi will solve this in the future it caused a lot of trouble over hear on an other project we just finished. and we worked insane hours just to get it done!!! :censored:

thanks for your reply folks
r



Replies: 0
avatar

news on this?



Replies: 0
avatar
  • Alex3D
  • Posted: 21 March 2009 05:50 AM

no.. i have the same problem



Replies: 0
avatar
  • sfu_Pete
  • Posted: 23 March 2009 11:35 AM

Yeah, I get it too.
But I found that it worked okay when I rendered it out of deadline.
I guess Deadline reloads xsi between each frame group, which forces the ram to be cleared out.
Maybe you could use something like that if it’s just a small thing? I think deadline is free for up to two machines…



Replies: 0
avatar
  • alenskas
  • Posted: 28 April 2009 09:53 PM

hi guys… i had the same problem. i was trying to render a large school of fish. using the ice cache node DOES NOT WORK. i think its cause it caches onlt the particle position..so when rendering the poor render farm is instancing the geometry each and every time. render pockets were 5-10 frames ! grrr
so then i dumped the ice cache node and used the simulate cache window from the environments tab in explorer. this worked !!
also plot geo and then loading to the mixer works well.

now im having a problem caching strands....it renders fine in region render, it renders fine with preview...but when i set the scene to render my strands stay behind and do not travle with the object the are linked to !!!!!!!!!!
huh ?



Replies: 0