The next issue we discovered matters not only for ABIT AN8 32X, but also for all other mainboards using passive heatpipe cooling solutions for the chipset. Of course, we are very pleased with manufacturer’s sincere intentions to eliminate the annoying noises generated by small high-speed chipset fans. However, when I was stress-testing the AMD Athlon X2 3800+ CPU overclocked to 2.8GHz with the S&M utility, μGuru issued a danger alert. The CPU temperature under the Tuniq Tower 120 cooler was absolutely normal, however the PWM Temperature sensor indicated that the voltage regulator components have heated up to 86-87oC while the normal temperature for them should not exceed 80oC.
Tuniq Tower 120 cooler is a highly efficient solution (you can read more about it in our review called Tuniq Tower 120 Cooler Review: All Super Coolers are Great, but Some Are Greater Than the Others), however its tower-shaped construction implies that the air is ousted towards the case fan or the PSU fan to be immediately removed from the system. In this case the heatsink covering the MOSFET transistors simply does not receive enough airflow. However once we replaced the Tuniq cooler with a Scythe Samurai Z, the situation got a totally unexpected turn. Instead of going down, the temperature of power elements reached 102oC and we had to interrupt the test. Looks like even the CPU coolers generating strong airflow in the nearby area are sometimes unsuitable for proper cooling of the power elements.
MOSFET transistors always heat up and are designed to withstand high temperatures. However, they can still get in trouble if they get overheated and work long enough in extreme conditions, like any other electronic components. Unfortunately, this is not the only concern we have. As we have already mentioned during the discussion of the PCB layout, the MOSFET heatsink and the chipset South Bridge heatsink are connected with a heatpipe. The excessive heat from the South Bridge should supposedly move along the heatpipe to a larger heatsink and get dissipated there. But the problem is that heatpipes work not in the desired directions, but according to common physics laws: depending on the temperature gradient. If the South Bridge heats up more transistors, the heat will move down the heatpipe away from it, but if it is not the hottest element, then the voltage regulator will add up to the South Bridge temperature!
To be fair I have to admit that these high temperatures were only noticed during dual-core AMD Athlpn X2 3800+ overclocking. When we tested a single-core AMD Athlon 3800+ the voltage regulator temperature didn’t exceed the acceptable value of 60oC. But did the South Bridge get any cooler in this case? I believe its temperature was also around the same 60oC.
I have always been very enthusiastic about the idea of passive cooling systems with heatpipes designed for mainboards. However, the current implementation of these systems cannot be called balanced or efficient. In certain situations these systems can do more harm than good to the platform by heating up the components instead of cooling them down. I believe the solution suggested by ASUS for their P5N32-SLI Deluxe and A8N32-SLI Deluxe mainboards is a very smart move in this case: they gave the users an opportunity to install additional fans on top of the passive heatsinks if necessary.
We didn’t conduct any performance tests for ABIT AN8 32X mainboard, because it will demonstrate the same performance level as any other Nvidia nForce4 SLI x16 based mainboard if tested in the same conditions. But if we try to achieve the highest overclocking results, then ABIT AN8 32X will inevitably lose because of the limited clock generator frequency and some memory restrictions mentioned above.