Slicing information of 3D volume

I’ve been trying to slice through a volume of information in an arbitrary direction and show the slices one by one - as described in 3D texture based volume rendering techniques.

I load my data which is 512x512x110 into a 3d texture and then proceed to render it onto a quad and show it using orthographic projection. This operation is pretty slow. It’s rendering at around 1 frame/second. Does that sound resonable or am I doing something wrong? What kind of performance do other people get when doing this? I believe the texture part is running in hardware, and I have a pretty cheap ATI card (X600 256 MB), how much better do you think it would perform with much better hardware?

Note, I’m not doing full volume rendering yet, just the slices. If there are any suggestions you have to optimize the process, I’d appreciate it.

Thanks for your replies.

110 is not a power of two.
X600 has hardware limitations related to non-power-of-two textures; if you have mipmap filtering on, or use certain wrap modes, it will fall back to software rasterization. 1 fps sounds like software rasterization.

Try using 512x512x128.

You’re right. The thing is flying now, what a difference. Such a silly mistake on my part.