Strategies to save/display low sample-rate data

I want to make waterfall displays of narrow bandwidth signals -- say +/-
20 Hz with FFT depth of 1024. Decimating to ~80 samples/second and
feeding that into a 1024 bin FFT is... not fast.

What's the best way to record the FFT vectors to speed up the display
for later analysis? What blocks would be used to read/write that data?

Are there other tricks for selecting a very narrow chunk of a wide
bandwidth signal to display at higher resolution at better speed?

Thanks!
John