Conclusions and Final Thoughts


I think I said it best in the original GTX 680 review: "I expected the GeForce GTX 680 to be faster than the Radeon HD 7970 from AMD but I honestly didn’t expect it to perform THIS WELL right out of the gate."  The same obviously holds true for the Galaxy version of the GTX 680 considering both cards are built on the same frame.  "In my testing, the GTX 680 2GB card was able to beat AMD’s flagship single-GPU card in Battlefield 3, Skyrim, DiRT 3 and Batman: Arkham City. The two offerings were pretty much a toss-up in Deus Ex and the AMD card came out ahead of the NVIDIA card in Metro 2033. "

There has been question and concern about the effect of GPU Boost technology and the variability of performance from one card to the next with the GTX 680.  While I have only tested 4 GPUs (two reference, Galaxy and one other for a pending review), the results have thus far fallen well within our standard error rates of 1-3%.  That still doesn’t mean that we won’t see bigger variances on a chip to chip level, just that with our small sample size we haven’t seen any indications of it quite yet. 


There are a host of new features included on Kepler, starting with the addition of being able to support more than two displays. Yes, the AMD cards can still support 6 outputs if you can find one of those magic DP hubs but I think that the four NVIDIA has included are probably enough for most users. I really still wish that NVIDIA wasn’t 2+ years behind on this — but we have it now so NVIDIA fans can stop being pestered by the AMD camp.

Interestingly, Galaxy was the only company to push forward on display outputs with Fermi, as we saw in our looks at the various Galaxy MDT graphics cards (GTX 570 and GT 520).  While we gave the company lots of credit for stepping outside the box and helping to push the NVIDIA market ahead, all of that work is likely lost now that NVIDIA has introduced the ability to natively support four displays.  

GPU Boost is the other big contributor to the success of Kepler as it enables the GPU to perform optimally for EACH game and allows the GPU clock to scale accordingly. In my testing the feature works — and works rather well — and yet still is flexible enough to allow gamers to overclock their new graphics cards with some easy to manipulate software.  I am a big fan of both the new Adaptive VSync and Frame Rate Target options as well, because they give users the ability and added flexibility that we haven’t seen before. The eternal debate of vsync on versus vsync off hasn’t been put completely to rest, but with the capability to smoothly scale under 60 FPS now an option on the GTX 680 I can see enabling that more and more in my own gaming.

Pricing and Availability

As of this writing, there are NO GTX 680s in stock at – but you should check for yourself as you are reading this.  The Galaxy model we are reviewing is also one of those that is missing in action and comes with a current price tag of $509.  You can also check Best Buy and Amazon for stock…but good luck.

NVIDIA claims that "demand is just insane right now" and that users looking to grab a GTX 680 should set stock notifications at places like  And while that is a grand idea for those of you with frequent access to your email and the Internet, it sucks for just about everyone that wants to buy a product that isn’t available.  Are we seeing limitations of the 28nm process technology from TSMC, yield issues on Kepler or a combination of both?  Chances are there are a lot of reasons for the lack of cards on the market, but I will give NVIDIA credit for sticking to the $499 price tag when they could have obviously charged $599 and still sold through their card stock.  


Final Thoughts

The Galaxy GeForce GTX 680 2GB is exactly what gamers are looking for today – a card based on the new Kepler architecture that includes all of the features introduced with the GPU including GPU Boost, frame rate target and adaptive Vsync.  If you can find one of these cards for sale you had best grab it quickly though as it likely won’t be there long. 

« PreviousNext »