You can speed up the first iteration, but after that there is NO similarity in the output so you can't re-use previous computations.

Also, as other posters pointed out, faster hashes are good for everything except password hashing, which is not the biggest use. In the case of password hashing, if the hash algorithm is twice as fast you can just run it twice as may times, so it doesn't hurt for it to be faster.