Power Consumption, Temperature, and Overclocking
Power Consumption and Temperature
All four cards tested feature the Asus DirectCU II cooling solutions. From experience these coolers are on much the same level as the MSI Twin Frozr units, but of course the Asus top end cards feature the triple slot solution rather than the dual slot of the MSI product. Just for grins I inserted the Twin Frozr III results for the R6970 and N580GTX XE for the temperature results.
In terms of power consumption the GTX 580 parts are leading the pack. The GTX570 comes in a ways behind. Interestingly enough, the HD 6970 runs slightly more efficient. The GTX560 is the coolest customer here in terms of draw. The MSI R6970 is clocked 60 MHz above stock as compared to the Asus 10 MHz, but it nearly matches the GTX580s.
With temperature we see some very nice numbers. The GTX570 with the full cooler but partially disabled chip is downright frigid when it comes to load temps. The EAH6970 from Asus is a bit toastier, but still well below 80 C. The MSI cards match the results from Asus, and in the case of these HD 6970 cards, the MSI card is 5C lower at load. The GTX560 has some very nice results as well, especially considering it has a significantly smaller cooler than the rest.
These boards all took to overclocking quite well. I am not the world’s most patient overclocker, so my results are pretty limited as compared to those that truly live and breathe this stuff. I used the GPU Tweak utility to adjust the clockspeed, but I left the voltage alone on these cards. Users who really want to push them as hard as they can should attempt these overclocks with a bit more power applied to the GPU. Considering the cooling abilities of these cards, they should be able to handle it. I also did not increase the fan speed from the auto setting. Again, people who are really pushing their systems will probably just set fan speed to 100% and go for it. Users just wanting a mild overclock and want to hear the game they are playing clearly will probably not want to set the fans at 100%. They get pretty loud.
The Asus ENGTX580 DCII compared to the MSI N580GTX XE Lightning. Both are big cards, but the Asus is just a hair shorter in length.
At stock voltage I was able to take the 560 Ti Top to just over 1 GHz. This is a nice overclock well above that of a stock 560 Ti. The extra 100 MHz over the already overclocked 900 MHz figure will give the user a pretty hefty boost in overall performance. The memory is not so lucky. Any NVIDIA card has problems over 4500 MHz effective, and that is due to the design of the memory controllers. All three of these cards were able to hit around 1140 MHz (4560 effective).
The GTX570 was able to hit 950 MHz without the voltage bump. People who have taken the leap and applied more voltage to this chip and go 100% on the fans have seen it go upwards of 1 GHz. That is pretty impressive for a card with this kind of performance in the first place. 200+ MHz above stock with the ease that I saw gave very significant results in gameplay performance. Noise from the fans do go up, even when set to auto. Still, this is a nice bump from the stock speed of this card.
When it comes to girth though, the Asus card is the one to beat.
The GTX580 did not fare as well as the others, but that can be expected due to the nature of it being a fully functional chip without portions disabled. That and every sample is going to be different. I was able to get to 920 MHz without tweaking the voltage. I would imagine that 950 MHz would be easy to get to if the voltage was raised a small amount. After that things get pretty toasty. This is again a significant overclock from the stock speed. Users wanting to push every ounce of performance into their games and benchmarks will likely take this route. Me? I’m pretty happy with stock speeds most of the time. Things tend to be more stable and work better.