Finally, the expected speed-ups from D-Wave, “~ 10⁸ times faster than an optimized implementation of QMC on a single core”.

Update: This is a bit embarrassing, I should have paid closer attention to the performance comparison graph (Fig. 4). There is indeed an exponential speed-up over simulated annealing but only a linear one over QMC. Also, see S. Aaronson’s article for more in-depth explanation of some of the potential problems with the interpretation of the results.