NVIDIA GTX 680 ReviewBosco , airman - March 22, 2012
» Discuss this article (308)
Airman's (Mike's) Thoughts:
I hope by now, after seeing these results, that everyone can come to an agreement with, or at least partially accept, NVIDIA’s claim that the GTX 680 is the fastest and most efficient GPU on the market. In nearly every test, the NVIDIA GTX 680 beats the AMD HD 7970. In some cases, the performance numbers of the GTX 680 actually top the dual-GPU card from NVIDIA. So not only is the NVIDIA GTX 680 the fastest single-GPU card on the market, it’s bordering the edge of the fastest GPU on the market, period! That’s a huge statement, but the numbers don’t lie. Not only does it have more juice than pretty much every other card out there, it does the amount of work that those cards can do and more, with less power consumption across the board. The Kepler architecture is clearly very impressive and I can’t wait for what’s next from NVIDIA.
Overclocking the NVIDIA GTX 680, as I mentioned at the beginning, is really no different than what us hardcore overclockers are used to. However, the card itself is smart about it on its own, even without the user’s interaction. The GPU Boost (out of the box) is really just a way for the card to vary its core clock in real time to achieve a constant power usage level. By being aware of its own power usage, it has the ability to ramp up its clock speeds to make up for the "performance opportunity" in a less power-hungry game. Many readers met this with little acceptance, due to the scare that the card may only overclock so far before its "hardware limitation" takes over and a wall is hit. I am happy to report that in all my testing, never did I have the card down-clock itself under load. I was able to maintain constant clock speeds throughout all my testing before the power target of 132% became a limiting factor. Rest assured that the GTX 680 is a mean overclocker. With core speeds of 1250MHz attainable by just about everyone out there, over 1300MHz for Dave and myself, the proof is readily available. I wish I could have gotten the memory clocked higher, but I still managed to break 6.5GHz on it! Not quite as high as what many HD 7970s can boast for memory clocks, but that’s not really what is important!
The Kepler architecture really is something special. It is a perfect representation of what we can continue to expect to see in the future. Core sizes are staying the same or even getting smaller, heat output and power requirements are decreasing, despite packing more and more transistors into a tiny little area, and the architecture is being optimized to really accomplish something magical. Managing to fit 1536 CUDA cores onto something smaller than 300mm2 boggles my mind, and don’t even tell me that there is one transistor in there for every two people on this world; 3.5 billion transistors! Hell, I’m already ready for the next big thing from either side!
With the GTX 680 and Kepler architecture in general, we get to see some cool new technologies, such as adaptive VSync and the current and upcoming anti-aliasing methods, FXAA and TXAA. FXAA is already available out of the box in some games, which is a much more efficient and even more effective AA method in comparison to MSAA (multisampling). TXAA is on the way, but it has been said to offer image quality similar to that of 8X MSAA with a load overhead of less than or equal to that of 2X MSAA. Adaptive VSync is a simple software implementation that still eliminates the tearing effects of frame rates higher than the refresh rate of your monitor, but it also reduces the stuttering effect from where it may have to drop from 60fps down to 30fps, 15fps, etc. Adaptive VSync essentially turns the VSync mode off when frame rates drop below 60fps, and back on above 60fps — resulting in a smooth transition as opposed to a hard switch from 60fps to 30fps.
I have long awaited the release of the GTX 680 for months since I first learned that it was on the way. I had a blast over the past couple of weeks learning more and more about the card and finally after having one in my hands. Seeing its results, I really don’t know how I was able to keep all this to myself. As we've even experienced at home here on OverclockersClub, there has been a lot of speculation and hearsay about what the GTX 680 is going to do and what it is capable of doing. Strong supporters of both camps have vehemently defended their own sides even while no hard evidence had surfaced, but hopefully agreements can now be met and everyone can be happy. To conclude, I have to say the GTX 680 is a great addition to the market and I can’t wait to see what the price of $499 does to the AMD cards. Price wars: right after this!
Bosco's (Dave's) Thoughts:
Ok, so where does Kepler leave us? After looking at our tests, you can tell NVIDIA has got the edge again in the benchmark wars. In some tests it was close, while other times you can see the GTX 680 pull ahead by decent margins. There is a ton of new features that can be talked about, but I will cover a few key things I think are worth noting:
Drivers - I will admit it's rather refreshing to have some decent drivers come launch time. Basically what I mean is everything works, no crashes, no stuttering, no flickering in games, nothing to complain about, unlike the last few driver revisions from AMD. I have said it a lot lately and I will say it again, if you don't have driver support for your products, raw performance is only going to get you so far.
Four Monitors - Four monitors supported on one card is a welcomed sight. Looking back at the 400 and 500 series cards, you needed two cards in order to run Surround, unless you had a GTX 590. You always had superior game play running SLI for the most part, but at an added cost. Now a single card can give you three monitors or even four if you want to go to that extreme — so again, a welcomed sight.
Price - Given the past track records from both camps, having the highest performing card usually commanded the highest price tag. Well not in this case! NVIDIA comes out of the gate with a MRSP of $499, which will be a huge surprise to a lot of people. First you pretty much beat your competitor in performance, power and features, then you beat them in price too. Ouch! From a buyers point of view, this is great news, as it may lead to the start of a price war. From a business stand point, you could see this as a smart move by NVIDIA to grab some market share.
Overclocking - My card managed to get a little higher than Mike's did toping out around 1355 on the Core. I fully expected cards to be different, but with a close distance, and for the most part that is the case here . But overall, 300+MHz overclocks on the GPU Core is quite impressive to say the least.
Features - Besides the GPU Boost, three other features I am looking forward to testing are the improved FXAA, TXAA, and Adaptive Vsync. In my opinion, from the few hours of testing I have been able to do on FXAA, some games do look better than MSAA, so it will be interesting spending a lot more time testing it out. Adaptive Vsync also seems to be a lot smoother than normal Vsync, but more testing will need to be done. Talking with NVIDIA you can now force FXAA in the NVIDIA Control panel for all games, looking forward to testing it out.
There is a lot of rumors that this card is actually a mid-range card and to expect more powerful cards later this year. If that is in fact the case, I for one can't wait to see how the rest of 2012 plays out. Overall, I am glad to see the launch of the GTX 680 and a chance to upgrade from GTX 500 cards without killing the bank.
- Very quiet at full speed compared to other cards
- NVIDIA Surround on a single GPU
- Power consumption values are excellent
- Overclocks very well
- 256-bit bus and only 2GB memory may be holding back performance