Lets say the processor accesses 2 bytes at a time, while accessing non-aligned memory which starts at position 1 rather than 0 in the underlying memory representation:
+----------------+
0 | |
+----------------+
1 | |
+----------------+
2 | |
+----------------+
3 | |
+----------------+
4 | |
+----------------+
What additional overhead the processor will have to bear (which wouldn't have to be done had this access been started from memory position 0 instead of 1)? Plus Is this overhead same if processor could access 4 bytes at a time?