speed with GL_FRAMEBUFFER_EXT

I am doing some offscreen texture rendering. The offcreen rendering consists of a single rectangle, however it takes 1/3 to 1/2 of the total frame rendering time to draw this rectangle.

The current gl rendering target is one of GL_COLOR_ATTACHMENTn_EXT, and a pixel shader is active when the drawing happens.

What can I do to speed this up? Is GL_FRAMEBUFFER_EXT just that much slower than the onscreen rendering? Are there other factors I should investigate? Perhaps someone knows the best set of options for the texture attributes?

I am testing on a GeForce FX 5200, will a newer nVidia card perform better?

Some questions:

1: What is the internal format of the render target?

2: You said the words “pixel shader” and “GeForce FX5200”. The performance of fragment programs under the FX5200 can be best described as a joke nVidia played on the gaming public. Depending on what your fragment program is doing (and what extension you’re using. The NV fragment program extensions have some additions that make it possible to optimize them for the performance limitations of FX hardware).

Originally posted by tagger:
I am doing some offscreen texture rendering. The offcreen rendering consists of a single rectangle, however it takes 1/3 to 1/2 of the total frame rendering time to draw this rectangle.
Just a sanity check, what do you expect? I mean, are you sure this is slower than it should be?
Is the ‘real’ rendering complex enough to take this relative measurement in real consideration?