Demented,
... not the calculation is the bottleneck, but the extruding-time of the volume ...
AFAIK the actual host-software does already patterning the inner volume - here you can win nearly 50 to 60 percent, as you extrude only every second or third line ...
But with 'cutting-only' the contour-outlines you can ignore the complete inner volume, what's thousandfold faster with micron-accuracy, as the amount of contour-lines rises only linear with the accuracy, but the volume-points quadratic - and as you reduce the slice-height with higher accuracy too, it's even a factory more!
So with a 'only contouring' fabbing i can drastically speed up the complete process or/and rise the accuracy until the actual limits ...
With knife-cutting there can be sheet-heights of some ten microns and line-accuracies in 0,01 microns too when you use an active rotating knife.
With diode-laser i have a spot-size of nearly 50 microns, so i can cut/melt something from 30 to 70 microns, energy-dependant.
With a powder- and ink-jet-system i can actually use powder-sizes of 10 to 30 microns, so with the inaccuracies with coagulating the powder-particles i'll have a finishing of smaller 0,1 mm - with infilling and micro-sanding there should be a possible surface-accuracy of maybe 50 microns ...
But with all methods i have an estimated accuracy of nearly 10 to 30 microns in Z and 20 to 50 microns in X,Y - so calculate with a 3D-pixel/voxel of 20x20x20 microns in comparison to actual output-size with a filament-thread of 500 microns - it's nearly a factor of x8.000 3D-points/voxels more with 'only' 20-fold higher accuracy!
Viktor