As a general rule, is it BETTER to create a complex mesh out of fewer
more complicated shapes or is it BETTER to use a larger number of
simple primitives? For example shape like -+-+-+- is it BETTER to
use a tube with three cylinders or one more complex lofted mesh?
where BETTER = faster render/display/less memory/etc.
A general rule of thumb would be to use the fewest number of
vertices/faces you can, but avoid using so few that you see distinct
linear edges in the profile of objects. The number of objects does
not appreciably affect rendering time, apparently, it's just the face
count. The Installation and Performance Guide probably has a
discussion of memory usage which you can check to see how the number
of vertices and faces affects memory usage.