Faster random uniform hash function

I need a little help in deciding the faster hash function for my program (in C++). Here is the scenario for which I need the better hash function. I have 4 million data and I would like to hash them [0, n] where n<<< 4M. The random hash function should be uniform e.g. if I hash 4000 data to [0,9], 400 elements should be hashed to 0, 400 to 2 and so on.