Query regarding CUFFT

I am using FFT for an Image Processing application. I am planning to replace my existing CPU based FFT (which is based on Cooley Tuckey algorithm) with CUFFT.

I have few questions regarding CUFFT.

( 1 ) Which is the FFT algorithm used internally by CUFFT. Is it Cooley Tuckey?
Then only I can compare the performance.

( 2 ) How much speed up can I expect for an Image of size 5k * 5k? A rough idea.

My CPU application takes around 71 seconds to complete.

Thanks in advance.

I think I have posted in a wrong discussion forum. I will this post this in CUDA Programming Forum. I request Forum delete this post from this discussion forum.

Sorry for the trouble.