Currently ChainIndex reads through the blk files in order to create a list of all of the blocks stored by the node before any further processing can begin. So far this process is single threaded, but it should be fairly trivial to make it multithreaded by reading each file in a separate thread since order does not matter.