You may get better results by leaving the clock speed settings alone and adjusting the long-term power limits to an acceptable level. That way you won't sacrifice peak performance (ie. latency, especially with only one or two cores in use) but will still get most of the power savings.
I'll echo the concern about the turboboost really helping at times. Often, you have a single-threaded process running that benefits from a single core running at a higher clock rate for a short period of time and improves the experience tremendously.
Without TB, I'm down to 8x 2.4, which is a bit sluggish.