help me understand new gpu arch please.

So this post should be taken all most completely as a question, as I am only repeating what I think I remember reading or what I believe in my limited knowledge. The 2900 creates 320 stream processers by having 5 shader units on 64 pipelines? 64x5=320. The 8800 series are 1 shader unit per pipeline but I thought I had read they can do 2 cycles per clock? Does this in effect make it 96x2 or 128x2? I am really unclear of any of that. The 2900 has a considerably thicker PCB. If that is true, does that in any way contribute to their cooling issues? If the 2900 is 64x5, can 3 shaders do pixel shading and 2 do vertex shading and send all that info down the pipeline in the same clock? I have a few more related questions but clearing this up would probably allow me to figure the rest out on my own. I (obviously) do not have the most technical grasp on this subject so replies that are not to technical would be appreciated.

Thank you everyone for your informative replies. I feel enlightened. Sorry- just a little bitter at the lack of help.