3 ms·
why is 7500 rounds the optimal number?
by dpweb 13y ago
why is 7500 rounds the optimal number?
- bdg 13y agoIt's not, PKCS5 v2 just makes a statement about doing as many as is tolerable to you, with 1000 being a minimum. http://tools.ietf.org/html/rfc2898#section-3 http://tools.ietf.org/html/rfc2898#section-3 IIRC, some hashing algos lose strength when this is done, but I can't remember what one. Also, I am not a Cryptographer.
- nly 13y agoBitcointalk is a busy site. It's could be the most they could get away with without burdening their server
- NKCSS 13y agoThey you just add a CUDA enabled vga card to the server and all of a sudden, you can handle a lot more load.
- nly 13y agoSure, and that's one of the problems with server-side key strengthening. You start off making a cracker do a thousand or million fold more work. They go and get better technology, like a GPU or an ASIC. So then you have to go upgrade your own hardware, or skimp and make the user wait for 1, 10 or 30 seconds while they login and you strengthen their password. It's an arms race and not something a star-topology (single server serving potentially thousands of users) is going to scale well on. You want those cycles rendering your HTML or doing something valuable. Client-side password stretching and memoization simply makes more sense.