Power, Heat, and Noise
We would expect the GeForce 8800 GTS 320MB parts to draw less power at idle and under load as compared to the 640MB part, as it does take more power to run more RAM. Some of our numbers support this, while others are different than we would expect.
The major power offender is the Sparkle card, but that is to be expected due to the power draw of the TEC attached to the GPU. Other cards' power differences can be explained by differences in the graphics BIOS (i.e. voltages and timings) on each card.
As for heat, what the Calibre lacks in power efficiency, it makes up for in temperature. This solution cools much better than the stock cooler attached to the 8800 GTS, which is nothing to sneeze at. The stock NVIDIA cooler is quite impressive at keeping hardware cool at lower noise levels, but Sparkle's setup wins the day here. With clock speeds similar to other cards, the TEC is clearly doing better when put under the same stress conditions, and we would expect end-user overclocking to be better than with stock cooling. It's unfortunate that Sparkle isn't willing to factory overclock their cards further, as that would make their offering much more attractive.
We should also mention that the ASC3 solution from EVGA didn't have that much of an impact on temperature. We do see slightly better than average cooling performance, but it's just not enough to sell us on it for cooling alone. We would once again submit that it looks cool even if it doesn't cool much better than the stock HSF.
Looking at noise pollution, all these cards gave us similar noise levels at idle. Under load, most cards again came in at about the same volume, but the Sparkle Calibre P880 LV kept very quiet as it cooled the hardware more efficiently than any other card we tested. We are certainly impressed with Sparkle's use of this cooling solution. While there are some drawbacks, the advantage is clear. This card will run much cooler than a simple air cooled solution.
We would expect the GeForce 8800 GTS 320MB parts to draw less power at idle and under load as compared to the 640MB part, as it does take more power to run more RAM. Some of our numbers support this, while others are different than we would expect.
The major power offender is the Sparkle card, but that is to be expected due to the power draw of the TEC attached to the GPU. Other cards' power differences can be explained by differences in the graphics BIOS (i.e. voltages and timings) on each card.
As for heat, what the Calibre lacks in power efficiency, it makes up for in temperature. This solution cools much better than the stock cooler attached to the 8800 GTS, which is nothing to sneeze at. The stock NVIDIA cooler is quite impressive at keeping hardware cool at lower noise levels, but Sparkle's setup wins the day here. With clock speeds similar to other cards, the TEC is clearly doing better when put under the same stress conditions, and we would expect end-user overclocking to be better than with stock cooling. It's unfortunate that Sparkle isn't willing to factory overclock their cards further, as that would make their offering much more attractive.
We should also mention that the ASC3 solution from EVGA didn't have that much of an impact on temperature. We do see slightly better than average cooling performance, but it's just not enough to sell us on it for cooling alone. We would once again submit that it looks cool even if it doesn't cool much better than the stock HSF.
Looking at noise pollution, all these cards gave us similar noise levels at idle. Under load, most cards again came in at about the same volume, but the Sparkle Calibre P880 LV kept very quiet as it cooled the hardware more efficiently than any other card we tested. We are certainly impressed with Sparkle's use of this cooling solution. While there are some drawbacks, the advantage is clear. This card will run much cooler than a simple air cooled solution.
Facebook
Twitter
RSS
24 Comments
View All Comments
gus6464 - Tuesday, March 27, 2007 - link
yeah same here, my first evga purchase was a 7800gt over a year ago and the next month the 7900gt came out so I sent my card to evga and they upgraded to a 7900gt at no extra charge. Shipping was fast and have had 0 problems with the card ever since. Now all I buy and recommend is evga.drebo - Tuesday, March 27, 2007 - link
From Page 12.
I don't quite agree with your conclusion here. Sure, the temperatures are only a few degrees cooler than the other stock cards, but the ACS3 card is also clocked higher than the rest of the cards. I think when you take that into account, the fact that the ACS3 solution cools better than nearly every other card is pretty impressive.
Other than some of the conclusions, very informative article. Definitely considering the eVGA card as an upgrade for my "aging" 7900GS KO.
RaistlinZ - Tuesday, March 27, 2007 - link
I usually enjoy AT's video card reviews, but this one seemed very rushed and with not much effort put into it at all. It reminded me of reviews you'd find from newbie review sites run by high-school kids.Sorry to say, but it's the truth. The whole "review" seemed like it only took you an hour to test the cards and post the review. :( Which is fine I guess, if that's what you were going for, but I expect more from Anandtech.
Anyone else feel this way?
yyrkoon - Tuesday, March 27, 2007 - link
You would probably be better served by waiting until nVidia irons out their drivers for these cards. However, I also must admit, I am tempted myself to buy one of these 8800GTS cards ( Titan Quest on my 7600GT does not seem to perform 'optimaly'). This being said, the games I play, for the most part play well enough, I think, for me to wait until nVidia gets their act together. Still, it IS tempting . . .