Making Themselves Heard: 11 GeForce4 Ti4400 and Ti4600 Cards

What's That Noise? It's Your Fans!

Until recently, an extensive cooling solution was considered a make-or-break feature of any card. After all, the right amount of cooling allows users to squeeze those last few MHz out of the chip. Today, cooling has become more of a marketing and design issue than one of performance, at least for graphics cards. Leadtek's flagship models are prime examples of this philosophy. Not only do they carry one heatsink each on the front and the back of the card - there are also two fans working in tandem to keep the chip cool.

Yet not everybody needs such an extensive cooling setup, and most will prefer a slightly less effective but much quieter solution. What good is the fastest gaming machine if you can't even hear the game dialogue or sounds over the din of those high-speed fans? So how much noise does a graphics card actually make? That's easy to discover: let's take some measurements!

That, however, is more easily said than done. Finding a meaningful way of measuring a card's noise level while still making the results comparable to others is a complicated, almost impossible task. For one thing, the cards are installed into a case with the fan facing down. For another, the distance at which the measurements are taken is of crucial importance, since the volume drops with each centimeter of distance. After much deliberation we decided on the following testing set-up: we put the card in a box with acoustic insulation and set up our measuring device 42cm away from the box and 10cm to the side.

Before every acoustic measurement, the voltage of each of the cards' fans was determined exactly. Although all fans are designed for 12V operation, this proved to be a mostly theoretical value. Gainward's card is the most extreme example. Although the fan header provides exactly 12V, this drops to a mere 8.6V when the fan actually starts spinning. Whether this is by design or by accident - the fan spins quite slowly and is consequently very quiet. When provided with the full 12V (from an external power supply) the card suddenly takes on the acoustic characteristics of a blow-dryer. The remaining candidates provided almost the exact required 12V, some of the Ti4400's were even right on target. The Ti4600 cards showed a spread ranging from 11.6V to 11.9V.

With these values established, we were ready to move on to the measurements themselves. This doesn't necessarily make the results meaningful though - they can only show us a trend. The two VisionTek cards illustrate this perfectly. Although both cards use the exact same fan and operate at identical voltages, the 4400 is a good 5db (A) louder, which results from a whirring sound not present with the 4600 card. We then took a look at the PNY card, also using the same fan, for comparison, and were surprised to find that it ranked somewhere between the two VisionTeks. This lead us to conclude that this family of fans is subject to varied quality. Lastly, let us reiterate that, despite our measurements, it isn't said that the fans actually run at the voltage we determined.. A variation of even +/- 0.1 volts can have a huge impact.

So was all this measuring and testing for nothing? Not quite. At the very least it should let us determine a trend. The subjectively quietest card of the group was ABIT's 4400 board, whose fan barely produced more than a murmur. It is followed by the PNY and the two VisionTek boards. Towards the middle of the field we find Gainward, ASUS, and Chaintech, whose card was characterized by a very unpleasant whirring sound. The second to last place is held by Leadtek's tandem-fans. The dubious honor of most noticeable card, in the negative sense, goes to MSI's entry. Producing almost 50db (A), this card can drown out even high-speed CPU fans.