Burst FFT: The burst FFT process one frame of data at a time. The core needs to finish processing and outputting that one frame of data before it can accept more inputs.
Buffered burst FFT: The buffered burst FFT takes a couple frames of data at a time. The core would need to finish processing those couple frame of data before it can accept more inputs.
Note: Buffered burst, burst and streaming FFT all use Block Floating Point algorithm.
Variable Streaming FFT: This is a streaming FFT. However, you can run-time configure the FFT size if you want. For example, if I have a 1024 point VS FFT, then I can change the FFT size on the fly (anywhere up to 1024 points). The change would apply to the subsequent packet.
Note: Variable streaming uses a radix 2^2 structure and it support fixed point (allows for natural bit growth) and floating point (IEEE 754 single point precision).
Also I recommend using the 8.0 version of FFT. Due to the interaction between the built in Avalon ST FIFO and the FFT core, the burst and buffered burst modes don't really behave correctly. (Take burst for example, rather than taking one frame of data, then core would have to take one frame + the next couple samples from the subsequent frame, before it would start computing). I heard that problem has been fixed in the 8.0 release...
I will post some examples later if I can find them on my machine...
Hope this helps...