Multiple windows

Maybe a stupid question … but I am thinking of writing a new app. But I would want something like 20 small double buffered opengl windows. With swap buffers and a 60hz refresh rate if I drew to them all I would only get max 3 ~ fps. What’s the best way of driving so many windows. A separate thread for each so they don’t block waiting for vsync ?

I would render to 20 different textures using FBOs, then draw each view as a textured quad, in one OS window. Then you only have to worry about 1 swapbuffers call, once all of the FBOs have finished rendering.

I think you can also use swap groups as well, but I’ve never used them (http://www.opengl.org/registry/specs/SGIX/swap_group.txt)

The ‘os’ windows I want to be physically separate inside the program. I could fake it somehow filling in the gaps but would prefer not to do that.

Okay I wrote the gui using multiple windows. Paints about 32 small windows with opengl, each with it’s own context. On my nvidia machine, the process usage is about 50meg for the whole app. On my ati machine I tested, was hd 7850 or something, the process ate about 1gig of memory. Anybody experienced this before ? I guess multiple opengl windows are not memory scalable with ATI drivers.

[QUOTE=dukey;1260641]On my nvidia machine, the process usage is about 50meg for the whole app.

On my ati machine I tested, was hd 7850 or something, the process ate about 1gig of memory.

Anybody experienced this before ?[/QUOTE]

Are you looking at virtual address space size, or physical memory consumption? Your ATI number makes it sound like the former.

In task manager the working memory set was pushing 1gig. I could see the memory increase x amount as i clicked on another tab to draw more windows.