|Zalman VF3000A VGA Cooler|
|Reviews - Featured Reviews: Cooling|
|Written by Bruce Normann|
|Friday, 10 September 2010|
Page 7 of 8
VGA Cooler Test Results
All temperatures are reported as measured by FurMark in degrees Celsius, and the ambient temperature for all tests was between 23C and 24C. Let's look at the results for idle temperatures first. The idle results reflect the lowest standard clocks that ATI runs on the Radeon HD 5xxx series of video cards. Under normal 2D conditions, the GPU loafs along at 157 MHz on the GPU core and 300 MHz on the GDDR5 memory. The core voltage is also reduced by each of the on-board PWM controllers at these low speeds. Here's a table showing how the two cards compare. The ASUS EAH5870 V2 idles at a higher voltage, slightly over 110 mV higher than the PowerColor PCS+ AX5870. At full tilt at the maximum 3D load, the difference between the two cards is much smaller, less than 15mV.
GPU Core Voltage at Varying Clock Rates
At idle, all three coolers keep the GPU below 35C, and there is only 2 degrees separating them. In the case of the stock coolers, I let the fan controller on the card adjust the fan speed. The PowerColor set the fan to 11%. Unfortunately, due to the simple 2-wire DC motor controller, I can't tell you the actual RPMs produced, only the percentages that were reported in GPU-Z. The ASUS controller set the blower wheel spinning to 21%, which registered 1000 RPM in GPU-Z. I set the Zalman fans manually, using the supplied Fanmate controller to the minimum setting, which yielded 1500 RPM.
The results are very close, are between 8 and 10 degrees C above ambient, and all three cooling solutions were essentially silent at these low fan speeds.
These load results are at the stock Radeon HD 5870 clock speeds of 850 MHZ on the GPU. For each card/cooler pair, the red bar with the higher temperature is the result with the fan running either on automatic, or at the minimum setting. The PowerColor fan controller pushed the speed up to 40% under load, the ASUS ran at 34% - 2200 RPM, and I kept the Zalman at the lowest setting on the Fanmate controller, which was 1500 RPM. At low speeds, the Zalman VF3000A beats out both stock cooling solutions, based on the strength of its large heat pipe/fin assembly, I have to assume. Fan noise was most noticeable on the ASUS, with its blower wheel construction, less so on the PowerColor single fan, and the Zalman made the least noise of the three, even with two fans running.
I like to run my GPUs a little cooler than that, and it was hard for me to sit there and watch them cook at low fan speeds, so I turned them up to 100% as soon as I could. Temperatures generally dropped by 10 degrees with the fans on high, and I breathed a little easier, especially for the ASUS card. If there is a downside, it was noise. The ASUS blower at 100% is spinning at 4350 RPM and is very loud and unpleasant, with a lot of higher frequency harmonics in the noise signature. The PowerColor fan was noticeable at full speed, but much less obnoxious, and the Zalman was again, the quietest of the three. At 100%, its two 92mm fans were spinning at 2700 RPM and were no louder than your average case fan.
If you're not unlucky, there is an extra 100 MHz of overclocking headroom available on the Radeon HD 5870 GPU. So, I took advantage of the situation to see how well each cooler handled 950 MHz of ATI's finest. With automatic (or low) fan speeds, the PowerColor and Zalman coolers each gave up an additional 3 degrees, with the Zalman VF3000A turning in the best performance, at 72C. This is despite the fact that the Zalman was still running at its lowest setting of 1500 RPM, and the controller on the PowerColor card bumped up the fan by 10%, to 44%. The ASUS fan controller kept the GPU at the exact same temperature as it was at 850 MHz, by raising the fan speed to 36% - 2370 RPM.
Turning the fans up to 100% again, gave some much needed relief on temps. This is really the acid test for these coolers, as this is where most serious gamers are going to want to run their cards at - maxed out. The PowerColor and Zalman coolers tied for temps, at 63C. The difference between them is noise, plus the fact that the VRM section was probably running cooler on the Zalman, with one of the two 92mm fans blowing directly down on that area of the board. The ASUS won the temperature test, but fails big time on the noise profile; at 4350 RPM, it's screaming like a salon-grade hair dryer. I turned it down to 70% to see how that worked, and it helped some. At 3850 RPM, it's still the loudest and most annoying of the three fan solutions, and the temperature of the ASUS rose to 62C at that speed. Temperature wise, this is roughly equivalent to the other two cooler's performance, but the other two still win convincingly on noise points.
To sum up, the Zalman matched the PowerColor PCS+ cooling solution on temperature, was noticeably quieter doing so, and delivered better cooling to the VRM section of the board. The downside to the aftermarket solution is the loss of automatic fan control. While I am not satisfied with the stock fan profiles of either card, they can be tweaked with software to behave exactly how I want them to. The manual controller that is supplied with the Zalman VF3000 also gives me complete control at all times (as long as I'm in the room), but I have to set and reset it by hand every time I want to make a change. Fortunately, the noise of the VF3000A is so low, I think I would just set it at 100% and forget about it. Cooler is always better in my book, and if I don't have to pay an acoustical penalty, why not keep the card as cool as possible, all the time? The ASUS is my choice if I absolutely want the best cooling performance, and the fact that I have run that particular card at GPU speeds over 1 GHz is a solid testament to that. I accept the fact that it makes a racket, even though I wish it didn't. If you want to verify that sentiment, take a look at the specs for the fan I have on my CPU cooler; that'll tell you everything you need to know.