Clock for clock overall dotham is aprox 90%~92% of a K8. It is faster in cache friendly applications (super pi for example), but for data intensive tasks (video, CAD) it lacks horsepower. A faster bus would surely help.
Remember, a dotham vs K8 exposes the weaknesses of the hammer line, which were well hidden against the inefficient P4: Integer operation and cache operation. However, against Dotham the weak SSE2 unit of the hammer looks potent. It would surely be a good battle.
Alex