Quote Originally Posted by SoulsCollective View Post
Surface area. That die is damn small and has no crenelations, fins, pits, grooves, anything to increase surface area. Although you are removing all those possible sources of inefficiency in thermal transfer, which should count for a pretty significant temp decrease, it's oversimplifying things to say that there is no possible way this could not work.
Except that it has worked - there is a thread floating around here that I saw a few days ago where someone in fact had already done this - his temps, if my memory serves me correctly, were mid-30s full load, over 4GHz. It just comes down to how much water is moving over the die - Over time, the surface area of water that is exposed to the die over, say, a minute, is astronomically higher than the surface area of the TIM/solder/IHS, which has a much higher thermal conductivity, but a static surface area.