I assume that if an specific mobile implementation of PowerVR MBX’s OpenGL ES 1.1 doesn’t provide any specific (and thus optimized) video memory but it just provides system memory for vertex data then we are not supposed to obtain any performance improvement by using Vertex Buffer Objects, but will it decrease performance or will perform just as vertex arrays do? Is it recommended in this scenario to use Vertex Buffer Objects anyway (for portability with future versions of the hardware or for other platforms that actually mount specific video memory)?
When I use Vertex Buffer Objects with static draw usage, every PowerVR MBX implementation still need to copy vertex data every frame?
You got the recommendation exactly right. On devices with PowerVR MBX you normally won’t get a performance improvement from using VBOs, but on PowerVR SGX you will.
However, if you have dynamic vertex data that is generated by the CPU every frame, don’t use VBOs on MBX.
While vertex data obviously needs to be read and processed every frame a copy is not generally necessary, but that depends on the system architecture.
Thank you Xmas! Then I’ll assume that for static vertex data I can use Vertex Buffer Objects without needing to worry about a decrease of performance even though there is no specific video memory.
Regards!
