It can denote rounding to the nearest penny, but it's not really a banking specific function (it's actually not widely used by banks as far as I know, but far more often in applications like digital signal processing, etc.), so it can also round to the nearest thousandths, ten thousandths, even dollars, for that matter.

The only difference between it and traditional rounding is the handling of values exactly halfway between two possible results, so if rounding to the nearest penny, banker's rounding would kick in with given a given value of 1.045, but not 1.044999999999999, nor 1.0450000000000001. It has to be exactly halfway, otherwise, just use the Round function and you should get the same results.

There is another difference, but it's more implementational than a directive of the function. Many of the functions available allow you to round things to the nearest nickel (values not denoted by a single decimal position), for instance, which isn't an intended purpose of traditional rounding, even though it's possible to do so. I think that is probably due to the name of the function as much as anything, although with the push to nickels (and their equivalents) being the smallest currency in many places, it has gained some traction.