To those debating heatsink heights: heatpipes.
I'm not really sure it's as simple as that. Heatpipes dont need to be vertical. In fact many heatsinks with built in heatpipes have horizontally mounted pipe configurations. Highend laptops are a good example. Even the little DC had a horizontally configured heatpipe. 8)
By going wider, you also need to go thicker. Otherwise, your temperature gradient will be too great at locations distant from the heat source. Once that happens, you can have square miles of heatsink and it will make nada difference in heat dissipation capability. The bulk temperature at distant areas of the heatsink would be too low to drive out any heat from the metal to air. The aluminum foil example presented earlier is a perfect demonstration of that. Surface area is good, but its the surface area that is very near to the heat source that does the bulk of the work. You can extend the heatsink perimeteral dimensions as far as you want, but very quickly it will yield only wasted metal, rather than increased heat dissipation. Most heatsink designs you find are designed to be the size that they are because that is the optimal size for that physical configuration (the videocard heatsink, for example).
That is true and the tinfoil example is good for the purpose of making a point, but it doesn't exactly mimic what's being proposed here because you're ignoring other factors that are
more important when you're not making a heatsink 1 mile long. When you double the length and width of a heatsink that is
2.5"x3" you are also increasing the
FIN surface area by 4 times. We're not talking about increasing the length and width by a mile and shaving off the fins so that it would look like a sheet of tin foil. Not only that, but we're using very thermally conductive alloys here so overall the net effect of not going taller
while only going a couple of inches wider is insignificant. You also need to factor in the chip size to heatsink contact area ratio too. In fact using that 1 mile heatsink example would actually make sense in a situatiion where you had millions of chips along the length of that 1 mile heatsink assuming the ratio of heatsink to chip area is appropriate. In my proposed solution, the total chip area (CPU+GPU+RAM) would be around 6 sq inches which goes very well with the 4x2.5x3" heatsink. That's a 1:5 ratio which is a far cry from that 1:1,000,000 "example".
You cannot arbitrarily increase dimensions w/o re-evaluating the base metal thickness, and assume it's just going to work "gooder".
And you cannot arbitrarily make up rediculous fantasy scenarios that don't apply to the specific implementation being proposed to make it "badder". We're not mounting a 30 sq in heatsink to just one chip, therefore the spread of heat through the heatsink isn't isolated to a single hotspot.
Someone earlier mentioned heatpipes as a solution, which could work to put more heatsink area to work, but even that can only be applied so far (no pun intended), and not without additional losses (and costs) in of itself, as well. The bottomline here is that, like most things, simply specifying "more" will not yield a workable solution. You got to re-evaluate the entire thing as a system to put "more" to good use.
If heatpipes are used in Revolution, it would be in a configuration like that used in highend notebook computers which do use heatpipes but still only take up about .75" of height. IMO, heatpipes might not even be needed at all.
