I'm using the machine in my sig, a 2.6GHz MacBook Pro. It's set to using 2 threads as it should be. I haven't set a custom number of threads. I found that upping the cache size from 400 to 1200 took 4 minutes off the render time for that benchmark.
I'll run some benchmarks tomorrow with Terragen 2.0 and 2.1 on Mac OS 10.6 and Windows 7 x64 and different cache sizes for comparison. That benchmark takes about 30 minutes to render, so all the different benchmarks will take a long time to do. If you have a good scene for benchmark, with water, clouds, and some objects for example, that might take less time to render, feel free to e-mail it my way.