subambaint insulation efforts will be the same regardless of wich method you use to get subambaint temps, so your equal there.
It depends on the purpose of your overclocking efforts.. If your wanting to benchmark dry ice is fairly cheap (decent container can be gotten for as little as $70 and dry ice can be anywere in the .25-just over $1 a pound) your load temperatures are going to be as low as a ok two stage cascade (not a great one)
Next stage IMHO for benching is a solid cascade (-100C load), could be as cheap as a couple of AC units and DIY time (and effort
), but to buy even a rotary DIY cascade they run around 1.5K.. The petty two stage cascades run 2-3K.
If your wanting something to run 24/7 clocks.. I would advise a solid single stage direct die, and if your GFX is not going to be the bottleneck sub-zero you could go chiller (and convert your WC loop for that)
have you dropped the multiplier so you could see how much FSB is in your motherboard?
And if you plan on going DX10 GFX.. just run as powerfull subambaint as you can on the CPU, DX10 cards are going to be CPU bottlenecked if the cards go subambaint (It apears that way and I've asked people that know a little better then I.. thats the answer I got, and if someone knows better post
)
you'll notice I kind of skipped TEC's.. if you want to utilize your WC loop a chiller would be more cost effective over time and much more effective temperature wise so I defalt to chiller when considering one or the other.