Computing cryptographic hashes of files on iOS and Mac OS X using the CommonCrypto APIs is fairly easy, but doing it in a way that minimizes memory consumption even with large files can be a little more difficult… The other day, I was reading what some people were saying about this on a forum about iPhone development, and they thought they found the trick, but they still had a growing memory footprint with large files because they forgot something fundamental about memory management in Cocoa. Even though they had a solution to read bytes from the file progressively instead of reading everything at once, it did not improve the memory consumption of their program when computing hashes of large files. The mistake they made is that the bytes read in the while loop were in an autoreleased instance of NSData. So, unless they create a local autorelease pool within the while loop, the memory will just accumulate, until the next autorelease pool is drained. But I think it would be very ineffi...