Question about modular reduction algorithm

[Cheung, Matthew G]

I was looking through the Algorithms section of the documentation to see if I could figure out what algorithm is used for modular reduction so I can compare the cost of reduction a 2048-bit integer by a 1024 bit integer vs reducing a 384-bit number by a 192-bit number.  Unfortunately I didn't exactly understand the Exact Remainder section.  Can anybody either help me figure out the difference in cost or point me towards different documentation that will clear up this issue?  Thanks.