Project and how it's going:
	Let's talk about rocket launches and NET dates!
	Real programs usually don't really go on schedule
	Some issues people are having:
		Large programs are complicated.  CASTER is no exception!
		There are problems CASTER solves (like zooming and moving)
		Layers are about 7.5 pixels thick
	Grading:  Have you all worked hard on this?

Since you're working on research with me:
	Mitochondria tracing
	New image volumes
	R15 grant getting and such
	Getting this into a tool that somebody could actually use
	My new project:  RTX bioinformatics

Neural Networks on a GPU:
	Let's review neural networks first
		Reminder:  Synapses and input weighting
		Grid?  Not in real life!  
		Backpropegation and training		
		"deep" learning
	Now that we know:
		Nodes:  How many?
		Nodes require large numbers of floating point calculations
		Each layer can be run with all nodes in parallel
		A tensor core can do 64 FMA operations per clock cycle
		640 tensor cores on V100, less on RTX (288 on 2070)
		Who is nVidia's competitor?
			Vega can perform 64-bit floating point and half 32-bit speed
		What does a V100 cost?
		What would it cost to beat one with Intel processors?
		https://software.intel.com/en-us/articles/intel-cpu-outperforms-nvidia-gpu-on-resnet-50-deep-learning-inference
		nVidia response:
		https://blogs.nvidia.com/blog/2019/05/21/intel-inference-nvidia-gpus/
	

Ok, trying something different with the GPU this time!
	Can we disappear some vesicles?