Ah, v2.00 runs a bit quicker than the standarized version 1.55.
v2.00 is actually intended for multiple socket/server environments. Version 1.55 is the standard used to compete globally.
Some people report the benchmark running quicker with the scenario you described in doubling or adjusting the thread count. Me, personally, when pushing up into the higher clocks, it actually hurt my times.
Also, if you run the benchmark several times, your times improve and you can pull down your best time. Usually between the 1st and 10th run in 32m.
What is the clock speeds at 9.82sec? Can you take a screen shot and post
