I apologize if I wasn't being clear with my post, I am aware that private keys cannot be computed from public keys (and the, er, *ahem* obvious security hole that would represent, lol). I imagine it would be possible to use the high level of parallel processing that a QC could be scaled up to to simply brute force private keys directly using the blockchain database. This would presumably take a QC with a high qubit count, but it's clearly the most common sense approach to hacking Bitcoin with quantum computing. It'd have to have a half decent rate of key discovery though, as the chances of finding a private key with alot of money in it's addresses could be pretty slim (this is an impression, I don't know any hard stats offhand, but I suspect the vast majority of keys have <10 BTC contained)
If you are talking about using a QC to simply brute force key 'collisions' in order to take the coins from random accounts, then QC is almost certainly not going to be the best way to do this. Moore's Law, assuming it holds up, would create a greater threat to the current algo first; IMHO.
Well, that sounds like pretty good news for Bitcoin then, it's going to take a change in the manufacturing process/transistor substrate before Moores law can carry on down past 10nm. So, in practical terms, Quantum Computing may never be able to brute force valid private keys, and an efficient transistor based solution needs at least a decades worth of more node process shrnikage as well as an economically viable new manufacturing technique. Go Bitcoin!