Simon Marlow writes: > I'm surprised if pointer access to memory is slower > than unsafeRead. You were right. Now that I have made some tests, the problem turned out to be elsewhere. Pointer access is not to blame. ;-) Peter