One of the points made in the discussions, is that the test took place with a Barton single core. Not sure where they got the information, but if it's true then it's indeed not representative for the latest generation top cpu's.
Whoever talked about the Barton core must have been drinking too much that night! Either that or they hated the block and wanted to slander AC. Here is what the author of that piece had to say about his test setup:
"So, the platform I used for the test that's shown was a C2D 6600 with machined top, a Thermalright backplate and Arctic Ceramique as TIM (normally I use Silmore, but I ran out of stock on it

)."
"The CPU was running at 3.833GHz with 1.46V, tested with Windows XP 64-Bit and the according version of Prime95."
"The loop I used had a PA 120.2, Laing DDC+ with Petra's top, Swissflow SF-800 flow meter, a small water filter made by Mips, t-line, and 7/16" Masterkleer with EK 1/2" barbs."
"I also used the Aqua Computer poweradjust to run the pump at 100% and 60% to simulate a lower flow loop (for people that like to use smaller tubing or a GPU block in their loop for example)."
"Water temperature and room temperature was recorded of course, throughout the testing the room temperature stayed within a range of ~2°C (I'd have to look that up, but the variation was small as all the testing was done in a row)."
Source is here.