Re: [Discuss-gnuradio] profiling the waterfall

[David Carr]

I spent some more time on the waterfall display.  Basically it runs 
about 6x faster now.  Major gains were had from precomputing the pens at 
startup and from removing unecessary code in the main loop.  A side 
effect of these changes is that the display can have an arbitrary 
mapping of value->color with no performance penalty.

There is still a bit of room for improvement, though.  We still need 
some sort of AGC or autoscaling.  Right now the range is simply fixed.  
Also, performance could be further increased and some new features such 
as window scaling could be added.

The new code is attached: (sorry if its a bad idea to send over the list)
-David Carr

waterfallsink.py
Description: application/python