Let's talk about run-time list building.
Hi,guys,have you ever thing about build list runtime.
In case that you are rendering a terrain which is a 1024*1024 map and the camera stays at a point,it is obvious that using a list is much better than send vertexs to card by glVertexPointer.However if glBuildList needs too mush time it would be a conflict whether to use it.
Perhaps you are thinking that using VBO would be a mush better way,but there is still a lot of old cards not supporting the new feather.
So what do you guys think about whether to use run-time list or not?
Actually, if you're playing games at all, chances are you've got the vertex buffer extension, ive lost my source, but iirc geforce2 and up will support it and radeon 9100 and up.
hope that helps
-Dan
hope that helps
-Dan
Yes,you're right,but there is strill a lot of old PC which have only embodied video card.and it is just a discussion for a possibility to use run-time list.Maybe it is very interesting to use it.
It is just for fun.
It is just for fun.
As long as you are creating the lists before the game starts, it shouldn't matter too much how long it takes, and the benefits are too good as long as the vertices don't move.
Vaguely recent cards that don't have VBO:
Radeon 7000
Any and all Sis onboard chipsets
So it's not a huge market without VBO. Every GeForce, as well as the Riva TNT and Riva TNT2 have VBO. Second, instead of trying to dynamically construct lists (which is a terrible idea), just use vertex arrays, compiled vertex arrays if available. Compiled vertex arrays actually perform very well; they basically just cost you a lot of bus bandwidth. (If you only need to construct a display list once ever, of course...)
Radeon 7000
Any and all Sis onboard chipsets
So it's not a huge market without VBO. Every GeForce, as well as the Riva TNT and Riva TNT2 have VBO. Second, instead of trying to dynamically construct lists (which is a terrible idea), just use vertex arrays, compiled vertex arrays if available. Compiled vertex arrays actually perform very well; they basically just cost you a lot of bus bandwidth. (If you only need to construct a display list once ever, of course...)
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement