Folks...consider the test system. i7 920 on an MCR220 with GPUs in the loop, if I read the post correctly. In this situation the loop is clearly rad-limited - there's simply insufficient surface area to effectively dissipate the heat being picked up from the components, so temps are crappy. With this hybrid block in the loop, you've essentially added in another radiator - thus the massive temp drop. If the blocks were to be re-tested fairly, you'd need to either control for the amount of additional surface area given by this hybrid block, or at least add many more rads, in so doing rendering the extra surface area given by the block a moot issue.

Of course, this may or may not be slightly missing the point, depending on how you view this - if space is an issue in your loop then a block such as this gives your loop more dissipation capacity and therefore better temps compared to a standard waterblock, even though a standard block is probably more efficient at removing heat from the CPU IHS. Borking the performance of the HK by lapping it also doesn't help the purpose of comparison.