Image filter demo so far:
	We can change the coloration!

Speedup:
	How much faster is the process with multiple cores?
	If we double the cores, and finish in half the time, that's perfect
		Speedup calculation:  This would be 2.0
	Perfect use of cores:  number of cores = speedup
	Less than perfect is more common

Embarrassingly Parallel:
	Our current image filter is this way
	Should have good speedup
		Could be less than perfect (shared cache, memory bandwidth, etc)

We need a task that takes more time!
	Can we outline the edges of things in a picture?
		Gradient over a neighborhood
		How about something from an mri?
		Membranes, and outlining

How many neighbors to consider?
	Ideally, this should be adjustable
		Pixels often have a fixed size, in nanometers or whatnot
		Often in the image metadata
		Biologists work in distance units, not pixels

A note about order:
	Do we change the image we're looking at?
		If so, we'll run into ordering problems!
	If not:  A fence
		When are all of the processes done?

Let's count how many pixels we consider part of an edge
	With a global count?  Locking and unlocking takes time

Just because the image is grayscale doesn't mean it can't have color added
	Easy way to outline things