- Articles & News
- For IT Pros
- Your Opinion
Without a decibel meter at my disposal, I'd estimate that the X1800 XL's stock cooler cranks out noise somewhere in the range of 45 dB(A) under low speed conditions, to 65 dB(A) when it's forced to spin at its maximum rate. That's only a subjective guesstimate, but what is a plain and simple fact is that the stock cooler is unacceptably loud, especially at full speed. At stock fan rates it is the loudest fan in our test system, and at 100% it is a leaf blower.
The VF1 Plus is rated by the manufacturer to output between 24 dB(A) and 32 dB(A), depending on the fan speed selected by the user. As mentioned above, 24 dB(A) is similar to what silent case fans will put out, which is darned quiet.
Regardless of the decibel numbers - which can sometimes be misleading because of the limited frequencies that the ear can hear - I can tell you from trying them both that the VF1 Plus is much quieter than a stock X1800 XL cooler. It was so quiet that it was difficult to hear over the case fans. The only way to know it was there at all was to listen for an oh-so-slight pitch change when changing the fan speed.
So how does a temperature reduction of 25°C translate to overclocking?
One of the advantages of the newer ATI X1800 and X1900 series graphics processors is that their voltage can be controlled in software. This allowed us to push the card much further, and to do it very easily compared to the days when you had to solder resistors on a card to increase voltage. To overclock, we used the AtiTool utility once again, which is helpful because it allows control of voltage, fan speed, processor and memory clock speed, as well as other settings.
With voltage tweaks in hand, we pushed the card as far as we could with both the stock cooling solution and the VF1 Plus:
With stock cooling, the All-In-Wonder X1800 XL was happy to overclock, but its temperatures would rise to astronomical levels so fast that we would have to shut it down. For example, the X1800 XL tested has a stock core voltage of 1.075 V. At a modest 1.2 V on the core, the card was more than willing to overclock from 500 to 650 MHz quite stably, but temperatures would quickly rise to 90°C under load. We chickened out and stopped the overclocking test, because this temperature is not acceptable to us.
The best we could do with stock cooling was 625 MHz core and 600 MHz memory clock speed. This still produced temperatures in the mid-80s under load; higher than we would like, but it remained stable. (The X1800 XL sample we tested on was an odd bird in this respect, as the core was happy to keep overclocking at very high temperatures without crashing. We have a suspicion that we could keep pushing this card until it melted, and it wouldn't complain about it until it simply boiled away.)
With the VF1 Plus, the temperature problems were completely eliminated, leaving the core's potential as the only limiting overclocking factor. We were able to increase voltage to 1.375 V, just 0.25 V higher than a stock cooled X1800 XT. At this voltage, the X1800 XL would run at 725 MHz. After a 3DMark05 run, the card's temperature was only 154°F (68°C).
Think about it: overclocked at more than 225 MHz above stock, and stressed with a 3DMark05 bench, the graphics processor was only 6°C warmer than the stock cooler at idle. As well, the processor was 22°C cooler than the card was when overclocked with the stock cooler - before we stopped the stock cooler test due to excessive temperatures.