Fan Noise

The ABIT AW9D-MAX is all passively cooled, so there is no fan noise. When the board is highly overclocked and the load on the system is changing, you can hear very faint electrical noises coming from the power circuitry of the board. The noises are not very loud, putting the board into a case should block them well enough.

Overclocking

Doing some basic overclocking is extremely easy with UGuru in the BIOS. The POST code display on the board and the extra button for power and reset make troubleshooting a failed overclock a breeze.

Even though there seems to be some kind of "overclock failed" detection, I had to reset the CMOS manually a few times to get the board to boot again. Another way is to physically change your CPU in the socket with another one, this will also remove all overclock settings.

Like all other i975 boards, the abit AW9D-MAX has a rather limited maximum FSB. When talking about "limited" here we're talking about 400+, some P965 board go into the high 500 MHz regions. Of course until Conroe such speeds were reserved to the most extreme of overclockers, but now you can go buy a cheap E6300 and crank it up to 500 MHz FSB or more.

With all voltages at default we could only reach a mere 373 MHz until the chipset started getting unstable. Based on what we saw on other sites, our sample seems to be one of the worse overclockers. By increasing the chipset MCH voltage, you can almost linearly increase the maximum FSB possible.


The graph illustrates how increasing the MCH voltage (X-Axis) increases the maximum FSB in the chipset (blue line, left Y-Axis).
What you can also see when you look past 1.85V MCH voltage, is that the blue line stops climbing as fast. This is because the chipset's temperature gets just too high, making it unstable. The chipset's temperature (red line, right Y-Axis) was measured with a temperature sensor sticked into the heatsink near the bottom.
My suggestion is that once you go beyond 1.80V in the BIOS you should start thinking about getting some airflow to that chipset heatsink. Without fan the increased temperatures may even make so much of a difference that your maximum overclock at 2.0V MCH is lower than at 1.80V.

As you see from the graph we got up to around 425 MHz FSB by raising the MCH voltage, without any additional cooling. With a fan blowing at the heatsink, at 2.0V, the maximum overclock was increased beyond 430 MHz.



If you want more, you have to do a very easy soldering modification on your board. ABIT's design gurus have not engineered their MCH voltage circuit 100% perfectly. There seems to be a thin trace or small resistor somewhere between MCH voltage regulator output and chipset. A thin trace has a higher electrical resistance than a thick one, imagine it as the electrons being so crowded together that not as many of them can move through the trace compared to if it was wider. The fix to this was pioneered by Hipro and involves soldering a thick cable to two points on the backside of the motherboard. Thick meaning something like the cables going from your PSU into the motherboard.



To verify that there actually is a voltage drop we connected a 6.5 digit Keithley lab DMM to each of the points and measure the voltages at different MCH voltage settings in the BIOS.



As you can see from the graph the drop is constant over the whole MCH setting range, about 0.07V. With a thick cable in place between the two points we finally got 444 MHz FSB out of the board, this required some cooling on the chipset heatsink to make it perfectly stable.

VCore Stability

Another interesting measurement is this VCore stability test. We used our DMM again to verify the CPU voltage as it gets fed into the CPU (measured at top left coil near the socket). The real voltages at idle and load are compared to what was set at that time in the UGuru software and in the BIOS. As you can see there is quite some difference to the setting in UGuru which I assume is a software bug. Between load and idle the differences are very small, ABIT has done a formidable job here getting the voltage regulators to work almost perfectly. If you take the BIOS VCore setting into account, the board undervolts between 0.008V and 0.05V.