Comparing arrays of floats?

[Straivers via Digitalmars-d-learn]

Forgive me if this has already been discussed, but how are arrays 
of floating point numbers compared in D? i.e. Is it a bit-by-bit 
comparison, is the std.math.approxEqual function get called for 
each element, or is it byte-by-byte across the entire array?

-Straivers