Gigabyte GTX 670 Windforce Graphics Card Overview
When Nvidia introduced Kepler in the form of the GeForce GTX 680 not long ago, it looked less like a launch and more like an onslaught. The GTX 680 really shook things up, certainly in terms of gaming horsepower, but also lower power consumption and temperatures. In this article, you will learn about Gigabyte GTX 670 Windforce Graphics Card in detail.
True enough, those are hallmarks we love to see, even expect to see nowadays. But Nvidia had a few tricks up the proverbial sleeve, with features such as Adaptive V-Sync and a new antialiasing mode.
Now, Nvidia is determined to continue that success with the launch of the GTX 670, the latest card in the Kepler arsenal. Sharing the same DNA as the flagship GTX 680, this new card is poised to offer the same innovations, but at $100 lower to be a bit more accessible to gamers who aren’t quite well-heeled enough for the Kepler elite.
Nvidia has unshackled the board partners with this launch, so custom designs are indeed hitting the shelves straightaway. One of those custom cards is the Gigabyte GTX 670 Windforce, a sleek behemoth that sports a triple-fanned cooler, increased Base Clock speeds, and a Boost Clock potential that appears to be second to none. Are some of those terms not ringing a bell? Not to worry, we’ll be examining each of them very closely today.
What you really need to know is this card runs cool like the underside of your pillow on a hot summer night. And fast enough to knock your current graphics card back into last year.
ASUS GeForce GTX also may help you to know more.
Closer Look Gigabyte GTX 670 Windforce
Specifications
- GPU Processing: GeForce GTX 670
- Base clock: 980 MHz
- Boost clock: 1058 MHz
- Process Technology: 28nm
- Memory Size: 2048MB
- Memory Type: GDDR5
- Card size: H=43 mm, L=275 mm, W=136.6 mm
As we saw with the GTX 680 launch, Kepler is a significant improvement in hardware technology, performance, features, and software from previous GeForce generation cards. As we’ll explore during the course of this review in further detail, these improvements are quite innovative, and really move the gaming world forward.
The GeForce GTX 670 is a bit of an interesting animal; it’s different from the GTX 680 in that the card will be almost exclusively available in non-reference designs. Nvidia has freed board partners to produce their own designs, so expect many custom models to hit the shelves for this launch. Consequently, there will be a large variety of clock speeds and specifications, more than we typically see.
The GTX 670 shares a very similar architecture to the GTX 680, and shares the same GK104 GPU. Further, the same 2GB of GGDR5 and memory clocks are also used, so on that basis alone we expect the GTX 670 to perform very well out of the gate.
In order to achieve this lofty performance, the GTX 670 comes with 1344 CUDA cores and 7 SMX units, while the memory has a 256-bit bus (four 64-bit). The stock base clock speed of the GTX 670 is 915MHz, with a typical boost clock of 980MHz. However, this Gigabyte GTX 670 Windforce card comes with a 980MHz Base Clock (core) and 15MHz memory. The GPU Boost clock on this Gigabyte card is 1162MHz, but this is the minimum you can expect to achieve.
Kepler certainly brings performance improvements, but the 28nm manufacturing process also reduces heat output. However, perhaps the most significant improvement is that Kepler brings twice the Performance Per Watt when compared to Fermi. “Awesome” might be considered an understatement in that context.
Lower power consumption appears to be a hallmark of Kepler, and that’s particularly encouraging. The GTX 670 has a TDP of about 170W or so. Killer performance is one thing, but killer performance at considerably lower power consumption is something special.
Let’s talk next about a groundbreaking new technology on Kepler cards: GPU Boost.
Overclocking
Historically speaking, a graphics card has certain clocks, and typically they have varied for idle and load states. And overclocking a card has involved raising the clocks, and perhaps voltage for stability, to achieve higher performance. Overclocked settings are then static, and in many instances, enthusiasts would even flash their card’s BIOS to “lock in” those settings.
All was well and good, but we craved something more, and Nvidia delivered: GPU Boost. We loathe using the term “game-changer”, but Nvidia did re-write the rulebook on that one, with the “old” way of thinking about clock speeds and overclocking, becoming anachronistic. Promising nothing short of a groundbreaking new direction in how a graphics card functions within a system, Nvidia brought the goods, and we were certainly impressed.
As we saw with the advent of GPU Boost in the GTX 680 launch, all bets are pretty much off with overclocking; it’s not the same anymore with Kepler. That continues here with the GTX 670. Since the clocks and voltages are not static, overclocking doesn’t actually result in “new” clocks per se; rather, overclocking Kepler results in higher overhead potential, or targets, with many different clocks and voltages possible.
Also keep in mind that GPU Boost provides a built-in turbo, which is essentially the same result as overclocking, except that it’s now dynamic instead of static. This means that you’re already getting a performance boost (ie-overclocked) card out of the box, so trying to manually increase the Boost Potential will not provide significant performance increases now. Essentially, manual overclocking isn’t as beneficial now because the returns are far less significant due to the built-in GPU Boost.
Not so fast. All I need to do is adjust one slider, maybe two, in EVGA’s Precision utility, in order to allow the card more overhead (ie-Boost Potential), and guess what? My OC beats your OC because my card is now running over 1200MHz with no problem.
Here’s the big kicker: my OC is dynamic. Yours isn’t. That means my Kepler’s clocks and voltages will automatically adjust on the fly, depending on the load required (or not). My OC will also downclock if the horsepower isn’t needed, while your OC is stuck at full throttle.
So my card is drawing less power, and running cooler and quieter, but can jump right back up to 1200+MHz in a fraction of a second if necessary. So who’s bleeding edge and who’s drinking the old lamesauce?
I’ve increased the GPU Clock Offset (ie-the Boost Potential) by +100MHz. Since my card can already do 1162MHz, this will get it up to 1262MHz. Remember though, this is a target speed, not a guarantee. I’ve also tweaked the Power Target slightly, increasing it to allow the card to increase the voltage automatically and account for the higher clocks.
What the power target needs to be is an educated guess, with perhaps a bit of trial and error if you’re really aggressive.
This is also why overclocking Kepler, while simple, provides diminishing returns because the GPU Boost is essentially already overclocking the card for you. You can certainly overclock the card, but you won’t be gaining too much horsepower beyond what’s already being offered for “free”.
The overall framerate improvements in games won’t be very much, depending on the Boost Clock and Boost Potential of your particular card.
So a quick performance bench run shows that we boosted our 3DMark 11 Performance score by almost 500 points with a couple of clicks of the mouse. That may not seem like much, but remember this Gigabyte card already has Base Clock and Boost Clock increases. The result below is on top of that already-overclocked state.
And on a side note for SLI users out there: what does this all mean if you’re running SLI? Well, it means that your two cards in SLI will indeed be running different clocks and voltages independently because each card has a different Boost Potential. It won’t be a terribly large difference, but it will exist. So keep that in mind if you’re monitoring clock speeds and performance in SLI; it’s entirely a new game now, and it’s good to remember this new approach for Kepler cards.
In terms of performance, this Gigabyte GTX 670 Windforce is an absolute behemoth, finishing within spitting distance of the stock GTX 680. Perhaps more interesting is that this card, Nvidia’s second-best, doesn’t beat AMD’s second-best card; it beats AMD’s best card. And quite handily as well.
That is impressive, no matter if you’re an Nvidia fanboy or not. Further, the performance per watt is a significant advancement with Kepler. Providing better framerates is one thing, but bringing more horsepower while sucking less juice is very impressive.
Summary
What began with Kepler’s GTX 680 now successfully continues here in the GTX 670. The simple fact right up front here is that the Kepler lineup is rolling along like a juggernaut with the launch of the GTX 670.
GPU Boost, in our opinion, is nothing short of revolutionary. The idea isn’t entirely new though, as Intel’s Sandy Bridge Turbo was the real innovator, but Nvidia appears to be standing on the shoulders of giants, as it were. Regardless, GPU Boost launched with the GTX 680 and it’s a resounding success here on the GTX 670 as well.
The latest GeForce launch continues Kepler’s evolutionary and revolutionary ideas and features, and the Gigabyte GTX 670 Windforce is indeed a force to be reckoned with, providing an outstanding product for gamers.