Multipass rendering slow on Radeon board

[zechampion]

I'm currently working on a multi (3) pass rendering algorithm that uses vertex (bound for 2 first passes) & fragment programs (pass 3) and glCopyTexSubimage().

This has always ran Ok on my GeForce 5600, but it lags on ATI Radeon 9600. Anyone has a idea?

Thanks.

[Humus]

Tried for instance reducing the size of the render target or similar to figure out what part is slow?

[zechampion]

The topic can be closed.

[zechampion]

I was stupidly calling a glTexImage2D 2 times/pass. The ATI OpenGL driver doesn't seem to check if texture size had changed, and hence if the memory allocation is necessary, while the NVidia driver does.