Hello Carsten,

let me explain why I try to use the MultiCore in the first place. What I 
have become aware of is that the scene graph entities are really memory 
expensive at run time. For instance a simple Node does cost, at least in 
a debug session, on windows roughly 480 byte (measured). In my case that 
is pretty much to much. I have a lot of very simple geometry. For 
reasons of picking, selection, highlighting and identification the 
geometries resides in their own scene graph branches. As my models are 
constantly getting bigger and bigger, I have to take steps to reduce the 
memory consumption of the graphical part of the application. So one 
first idea was to use the MultiCore core.

Unfortunately, that does not solve my problem, as it turned out that the 
Shader based material definitions are quite expensive so that they 
overshadow all attempts with the MultiCore objects.

What is the expected memory footprint of a Node, MaterialChunk or 
ShaderVariableVec3f?

My measurements yields in debug mode on Windows x64 (in byte):

Node = 484
ChunkMaterial = 456
ShaderProgramVariable = 640
ShaderVariableVec3f = 616
SimpleAttachment = 480
MaterialGroup = 432

Maybe I did something completely wrong with my measurement. What I 
actually did was to stop at the respective code line with my debugger 
and setting a break point into the memory manager's allocator function. 
Then I wrote down the requested memory sizes and summed them up.


How can I reduce the memory footprint beside of sharing materials?

How do others work with huge models? I.e. is it state of the art to 
build the scene graph up from 'ChunkMaterials' and 'ShaderVariables'?

Do other people have had similar problems and how do you solve them?

Best,
Johannes



------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
Opensg-users mailing list
Opensg-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to