Distributed OGRE

[pgavigan]

I'm working on an application that will involve lot's of networking (and involving OGRE of coarse). It would be really cool if OGRE had built in functionality for distributed application, on an array of computers over a network. I'd be curious to hear what others think about this.

[Game_Ender]

I too think it would be cool if there was a more integrated library to help me create my programs, but what core feature of graphics rendering engine requires distributed computing?

I think it would be cool if someone created a plugin that provided support for using Ogre in a Cave type environment where each screen is driven by a different computer but is managed like a single application. That just seems to specific to be put into the core though.

[Kencho]

I only see the strength in scene management, as most of the rendering power goes to the gfx card after all. I think an intermediate layer somewhere around the scene management layer would be the only benefited from that. Determine visible/invisible elements and such, in a really large environment, is the only benefit I can think of right now. Besides that, it would be application-specific and not a graphics engine issue at all.

[Sethos]

This is an interesting idea. A distributed scenemanager, which syncronizes the scene hierarchy from several computers. If just I had a Cave at home

[Falagard]

It would be really cool if OGRE had built in functionality for distributed application, on an array of computers over a network. I'd be curious to hear what others think about this.

That's way beyond what Ogre's focus is, which is just rendering.

Doesn't really make any sense.

[pgavigan]

This is an interesting idea. A distributed scenemanager, which syncronizes the scene hierarchy from several computers. If just I had a Cave at home

I'm building a cave... part of why I bring it up! I'm planning on passing it commands using a network, but also binding it in with larger scale simulations.

[Game_Ender]

You might consider a distributed object system like Ice. You could make Ice interface for Ogre's generic SceneManager interface (and a decent portion of Ogre in the process). Then with enough design you should be able to just propagate the calls that change scene state to the remote scene managers and keep all queries to the local one.

[pgavigan]

Another thing to consider too is the idea of distributed computing between processor cores and multiprocessors, not just between computers over a network.

[Game_Ender]

For that you should really have a look at the many threads about multithreading and Ogre. In short- There is not a lot to parallelize in Ogre and most of the area that was is already parallelized (Resource Loading).