Being powers of 2 is more efficient because computer memory/buses/caches are aligned to those sizes. It does not necessarily imply that they are less secure tha

Being powers of 2 is more efficient because computer memory/buses/caches are aligned to those sizes. It does not necessarily imply that they are less secure than with fields of non-power-of-2 sizes. There might even be a security benefit to having these canonical sizes because it simplifies implementation and makes it easier to perform constant-time operations across different hardware. If you consider that some hardware has additional costs to reading non-aligned memory, you open up the possibility for timing-attacks and other implementation bugs brought about by additional complexity. Sometimes simple is better.