Blogalog 1.0.3: qubits v. quantum gates
There are a ton of interesting thoughts in Wook's latest posting. I think we'll leave the bigger issues of quantum architecture for a later posting, though I'm looking forward to that. In the meantime I'll encourage you to Google on "quantum Turing machine" and noodle on which parts of the TM need to be quantum. For some concrete stuff on where we are with this, see the references from my Aqua recommendations page and my arithmetic page.
One bit of terminology we ought to straighten out. There are two types of qubits, "flying qubits" and static ones. Flying qubits are typically something like photons. They pass through devices that cause computation, something like data values pass through gates in a classical circuit. Many quantum computing technologies use static ones, such as quantum dots or the spin of atomic nuclei. In that case, a qubit is like a bit in a register, and the "gates" (often microwave pulses) feel more like instructions to a classical computer architect. The "program" is usually called a quantum circuit in either case.
So, rather than saying that a PDP-8 takes 10k gates for its CPU, the question is, how many bits of storage does it have? A machine that can run a program might want, say, a few kilobytes.
In most quantum computing proposals, an underlying assumption is that all qubits are equal; any qubit storage location can be manipulated in an arbitrary way. It's like they are all bits in a classical register; there is no RAM/register distinction and no "load" instruction. One of the few proposals that separates qubit storage from "action" locations is the scalable ion trap processor from the Wineland group at NIST. See http://qubit.nist.gov/ and the Kielpinski Nature paper. We in the Itoh group have been thinking about how to combine different storage technologies into a larger device, like the difference between cache and RAM. Haven't gotten very far yet.
This has already gotten long, so I'll leave simulation and the qubit flythrough for another posting.