Thanks!
Thermodynamics pretty much working now. The image below shows locations of mantle plumes (superimposed black dots) and a map of the crust, blue = cold, red = hot, each pixel = column of goxels, with an overly dramatic amount of heating going on. The crust is currently stationary. The temperatures you see are actually being filtered through conductiveness, specific heat, and density all properly by Fourier's law, with just a couple shortcuts (causing the minor and insignificant banding). Heating follows Gaussian, not linear falloff too! And since the plumes are all equally active for now, reads from a dynamically stored table of circle distances.
For a thin ocean crust, this algorithm goes quickly enough on a 300x300 map that I can't even snap a screenshot quickly enough before it gets too hot without adding in pause functions.
Next step is cracking the crust along the weaker hot paths (pathfinding with temperature as cost, for now), to form plates, then determining Voronoi maps of nearest plume neighbors for calculating plate movement vectors, and filling in new ocean floor and volcanism where stuff collides.
For now, volcanism is just a number stored in the column so I can show it in the images. Later will mean actual volcanism and trigger new algorithms, etc.
This isn't just a simple image of only thermodynamics. The coordinate system is already in place to handle all the wrapping of coordinates, and all set up to be ready to handle plates moving independently too (they're stored in separate vector lists of columns, so I only have to chnage one offset number to "move" the plate basically). System already loads up all the goxels as basalt, and has 3D structure, etc.
Getting there!

Top left panel is the actual map. Rest of it is copied/pasted to show how it wraps around correctly.