SLI or Crossfire? GTX680 or HD7970? 1, 2, 3 or 4 cards? Who is number 1? Join us as we pitch the most powerful setups from AMD and NVIDIA against each other. ..
Product on Review: KFA2 GTX680 2GB GDDR5 (QUAD SLI)
Manufacturer: KFA2
Sponsor: Overclockers UK
Street Price: £443.99
Like most enthusiasts, especially those that either bench or game at the high end, what we all want to know is ‘who is best’? In years gone by it was a simple affair of pitching one card against another however with multiple GPU setups now common place there is more to evaluating a graphics cards performance. No longer is it mano-a-mano, 1 vs. 1 because to truly examine the performance of a high end graphics card you also have to look at how it performs when in tandem with another. SLI and CrossfireX are critically important when faced with the decision of what to buy because it’s all very well buying the fastest single card, if it scales poorly in multi-GPU configuration you could be disappointed if you wish to expand on your GPU setup.
So when our friends at Overclockers UK sent us KFA2’s latest Kepler equipped graphics card, the mighty GTX680, we were intrigued to see what it could offer over other GTX680’s. More so because they were sending not one but four of these monster cards for us to analyse. Couple this with the fact we happened to have a few HD7970’s hanging around the office and we were ready to pitch an NVIDIA army against the forces AMD. Fun times ahead!
We were impressed with KFA2’s GTX570 with it’s funky cooling design. Sadly, the product we have for review today is pretty much reference by design but that does not mean it’s a slouch by any stretch of the imagination – far from it! Linking four of them together poses some interesting questions:
How do they scale? What is the power consumption like? Are they noisy? Are they hot? And of course the question everyone wants to know – how do they compare. So for this review we will be looking closely at both cards and comparing 1v1, 2v2, 3v3 and 4v4 which will allow you to make the best choice dependent on your preference of multi GPU setup.
About KFA2
KFA2 exists for 2 main reasons:
1. To bring the best quality and performing products to market
2. KFA2 is going to “Kick Friggin Ass” and take names later!
KFA2 are a sole European focused brand, providing European based support, warehousing and RMA facilities. Our primary goal is to provide products based upon two key elements “Performance and Quality” with peace of mind, knowing that all support related services are located within Europe.
KFA2 exists for 2 main reasons:
1. To bring the best quality and performing products to market
2. KFA2 is going to “Kick Friggin Ass” and take names later!
KFA2 are a sole European focused brand, providing European based support, warehousing and RMA facilities. Our primary goal is to provide products based upon two key elements “Performance and Quality” with peace of mind, knowing that all support related services are located within Europe.
Let’s see how the specifications compare…
Specifications (GTX680)
Firstly, comparing the GTX680 with it’s forbear, the GTX580 we see that there are some significant changes. Immediately identifiable are the amount of stream processors which have tripled. from 512 to 1536. The ROPs have however dropped 25% but to compensate the core and memory clockspeeds have been massively increased, with the GTX680 being the clocked slightly above the 1GHz mark.
The battle everyone wants to see though is the GTX680 vs HD7970 and comparing the two cards against one another is a good place to start. On paper at least, the GTX680 appears to have the slight upper hand. Sacrificing a gigabyte of GDDR5 and running with the 256-bit interface, the GTX680 cannot hope to compete in total bandwidth available losing out to both the HD7970 and HD7950 by a fair margin despite having a much higher memory overclock. The tables are however turned when we study the texture fillrate where the GTX680 comes out on top.
It was almost two years to the day that we saw the introduction of NVIDIA’s Fermi, NVIDIA’s previous architectural change. The GF100 featured 512 stream processors in a 16×32 format built by TSMC in a 40nm process. Sporting OpenGL 4.0 and DirectX 11, enthusiasts climbed over each other to get the latest and greatest GPU from NVIDIA. While the performance was there, power efficiency was very poor with the cards consuming lots of power and running very hot.
Enter the 5 series 8 months later and many would say the GTX580 was the card the GTX480 should have been. It was more powerful thanks to the fully unlocked memory controller (6x64bit), ran cooler and thus became a firm favourite among the high end GPU market.
For a little over a year the GTX580 enjoyed sitting atop of the GPU pile (save for dual GPU variants) and it wasn’t until the AMD HD7970 emerged in December 2011 that the GTX580 lost it’s performance crown. NVIDIA however had not been resting on their laurels…
GTX680 GPU
The successor to Fermi is codenamed Kepler. Listening to feedback of Fermi from gamers around the world, NVIDIA sought to create a graphics card that was not only the best money could buy in terms of performance but crucially it had to run cool and quiet to make for a power efficient model. Kepler addresses these issues in two ways, GPU Boost which we will discuss further in our overclocking section of the review and a redesigned streaming multiprocessor.
Kepler Core
The GTX680 is equipped with a GK104 core which NVIDIA claim is their highest and most power efficient GPU to date. Fabricated on the 28nm process, every component was designed with power efficiency in mind to ensure the GPU gave the best performance-per-watt possible.
While the 28nm manufacturing process holds the bulk of the power saving features, the way the new architecture works is also key to furthering Kepler’s power reduction. SMX, an evolution of the SM from Fermi is NVIDIA’s new streaming multiprocessor:
The SMX now runs at the graphics closk speed rather than 2x that speed as previously but because the GK104 has 1536 CUDA cores (8xSMX), over Fermi’s (GF110) 512 CUDA cores (16xSM), Kepler allows the GTX680 to operate at twice the performance per watt measurement compared to the GTX580.
The clock throughput of FMA2, SFU and texture operations have all been significantly increased and while some operations still retain the same speed as the GTX580, the GTX680’s much higher core clockspeed ensures a substantial increase for all GPU operations.
To feed the new SMX, each unit has four warp schedulers, each capable of firing off two instructions per warp, every clock. With a redesigned scheduling function, the GTX680 is a lot less complex and much more streamlined allowing the compiler to determine which instructions will be issued and can provide this information direct to the hardware block rather than going around the houses using Multi-port decode and a register scoreboard along with the dependency check before the information gets issued. This method is both quicker and more importantly, more power efficient.
Perhaps the biggest change to the processor core is the omission of the old Shader Clock. The shader clock was getting tired as it was part of the old Tesla architecture which was implemented as an area optimisation. Because of the way Kepler executes instructions at a much more streamlined manner and at a higher clock rate, fewer copies of the execution unit need be made.
Another improvement over Fermi with the introduction of SMX is the redesigned Polymorph engine (2.0). The Polymorph engine takes card of the tessellation workload of the GTX680. The design of the new Polymorph engine is to ensure that the tessellation we see on screen has little (as possible) impact on the rendering performance. What you may find bizarre though is that NVIDIA have cut the Polymorph engine count in half to 8 from Fermi’s 16 yet claim it is almost twice as quick. This is thanks to the improved core an memory clockpeed.
One area where NVIDIA are shouting from the rooftops is the power consumption area. With a max power draw of 195W compared to 250W of the HD7970, the GTX680 should save a few more yards of the Brazillian rainforests (and of course cut your electricity bill) than the HD7970. Price wise there is little to choose between them. The GTX680 is a little more expensive yet only because of the recent price drops of the HD7900 series. At launch, the HD7970 was significantly more expensive.
Overall, looking at the specifications you would be forgiven for thinking that Kepler is not the GPU everyone was hoping for. Sure it has faster clockspeeds but how does that transpire to performance when the core hardware (stream processor count, memory size) is not as impressive as AMD’s?
Let’s take a look at the competition…
Specification (AMD HD7970)
The HD7900 series features the new Tahiti core which is the successor to the Caymen core of the HD6900 family. The Caymen core was based on a 40nm manufacturing process so it is fair to say that perhaps the biggest improvement to the Tahiti core is the reduction to 28nm process. This has enabled AMD to cram 4.31 billion (4,312,711,873) transistors into the new HD7970 core as opposed to the 2.64 billion on the HD6970 Caymen XT.
Comprising of what AMD calls ‘Graphics Core Next (GCN)’ archetecture which is optimised for heterogeneous computing, the new Tahiti core has upto 32 CU’s (compute units) with dual geometry engines along with 8 back-ends comprising of 32 colour ROPs per clock and 128 Z/Stencil ROPs per clock. Each CU has a Vector unit and Scaler Co-Processor and can execute instructions from multiple kernels.
The VLIW4 SIMD has evolved to the GCN quad SIMD which equates to similar computational power yet grants better performance, threading and simplification. The GCN also has a new Cache hierarchy with 16KB instruction Cache and32KB Scaler Cache shared between 4 CU’s. The L2 Cache is accessible by each CU with it’s own separate process to ensure coherency. These changes increase the GFLOPs by 140%.
Tesselation on the Tahiti core is also enhanced by increased Vertex re-use off-core which theoretically at least, means the 7900 series has 4 times the throughput of the 69XX series. The are improved ansioscopic filtering algorithms too which come at no performance cost allegedly. These new algorithm improvement will allow the use of Forward Rendering with 1000’s of potential light sources and will also mean MSAA can be rendered easily.
Looking at the schematics above, we can identify the bulk of the GPU is taken up with the 32 CU’s. Above these are the geometry engines. To the left and right are the 8 Render back-ends, 4 each side. There’s 768KB of L2 cache along with 6x64bit dual channel memory controllers making a 384-bit GDDR5 memory interface capable of delivering up to 264GB/sec. All of technology coupled together is transferred through a fully compatible PCI Express 3.0 x16 bus interface. Have no fear though because the cards are backwards compatible with PCIe 2.0.
Comparing the HD7970 against its competitor, the GTX680, the AMD card has the advantage of more stream processors but because it is lower clocked, the GTX680 has a higher texture fillrate. Memory size goes to the AMD card while memory speed favours the NVIDIA counterpart. The pendulum swings back to AMD card when total bandwidth is considered though with the AMD card gaining a 37% advantage. The NVIDIA card however consumes 20% less power (load) than the HD7970 so while the NVIDIA care may present a higher initial outlay, over time it may prove to be the cheaper card, depending on how long you plan to keep it.
Let’s move on to the features of both cards…
Features
NVIDIA GTX680
DirectX 11
The first fully Microsoft DirectX® 11-compatible GPUs in their class, the ATI Radeon™ HD 5000 Series are designed to deliver unbelievable HD gaming performance for the latest games and DirectX® 11 applications so you can dominate the competition.
HDMI Support
MSI graphics cards offer the HDMI functionality to provide access to premium digital content. HDMI enables gaming enthusiasts to seamlessly connect to their LCD monitor to bring the ultimate multimedia experience. (Only 1 single HDMI cable is required while streaming audio/video data to flat-panel display devices, such as Plasma/LCD TVs, or projector.)
NVIDIA 3D Vision
NVIDIA® 3D Vision™ technology is a combination of Graphics Processor Units (GPU), specialized 3D glasses, software, and certified displays and projectors that deliver an immersive 3D experience on your PC or workstation. For a complete list of system requirements for NVIDIA 3D Vision Technology
NVIDIA 3D Vision Surround
NVIDIA 3D Vision Surround expands your field of view across three 3D monitors for Breathtaking panoramic gaming. At an ultra-wide resolution of 5760×1080, games like Battlefield Bad Company 2 have to be played to be believed.
CUDA
CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU to speed up the most demanding tasks you can run on your PC.
SLI
NVIDIA SLI technology links multiple GeForce GPUs together for dramatically improved graphics performance. With over 1000 supported applications and used by over 94% of multi-GPU PCs on Steam, SLI is the technology of choice for gamers who demand the very best.
NVIDIA PureVideo
Blu-ray and HD DVD movies are bringing a new level of movie-viewing experience; with high-definition image quality far surpassing standard-definition DVDs. NVIDIA PureVideo™ HD2 technology provides unsurpassed Blu-ray and HD DVD movie picture quality.
Dual Link DVI
Dual-link DVI convertible output supports Video/Audio signal integration through DVI to HDMI interface, which drivers resolution of a digital display up to 1920×1200 or higher.
PhysiX
PhysX™ is designed specifically for hardware acceleration by powerful processors with hundreds of cores. Combined with the tremendous parallel processing capability of the GPU, PhysX™ will provide an exponential increase in physics processing power and will take gaming to a new level delivering rich, immersive physical gaming environments
The first fully Microsoft DirectX® 11-compatible GPUs in their class, the ATI Radeon™ HD 5000 Series are designed to deliver unbelievable HD gaming performance for the latest games and DirectX® 11 applications so you can dominate the competition.
HDMI Support
MSI graphics cards offer the HDMI functionality to provide access to premium digital content. HDMI enables gaming enthusiasts to seamlessly connect to their LCD monitor to bring the ultimate multimedia experience. (Only 1 single HDMI cable is required while streaming audio/video data to flat-panel display devices, such as Plasma/LCD TVs, or projector.)
NVIDIA 3D Vision
NVIDIA® 3D Vision™ technology is a combination of Graphics Processor Units (GPU), specialized 3D glasses, software, and certified displays and projectors that deliver an immersive 3D experience on your PC or workstation. For a complete list of system requirements for NVIDIA 3D Vision Technology
NVIDIA 3D Vision Surround
NVIDIA 3D Vision Surround expands your field of view across three 3D monitors for Breathtaking panoramic gaming. At an ultra-wide resolution of 5760×1080, games like Battlefield Bad Company 2 have to be played to be believed.
CUDA
CUDA is NVIDIA’s parallel computing architecture that enables dramatic increases in computing performance by harnessing the power of the GPU to speed up the most demanding tasks you can run on your PC.
SLI
NVIDIA SLI technology links multiple GeForce GPUs together for dramatically improved graphics performance. With over 1000 supported applications and used by over 94% of multi-GPU PCs on Steam, SLI is the technology of choice for gamers who demand the very best.
NVIDIA PureVideo
Blu-ray and HD DVD movies are bringing a new level of movie-viewing experience; with high-definition image quality far surpassing standard-definition DVDs. NVIDIA PureVideo™ HD2 technology provides unsurpassed Blu-ray and HD DVD movie picture quality.
Dual Link DVI
Dual-link DVI convertible output supports Video/Audio signal integration through DVI to HDMI interface, which drivers resolution of a digital display up to 1920×1200 or higher.
PhysiX
PhysX™ is designed specifically for hardware acceleration by powerful processors with hundreds of cores. Combined with the tremendous parallel processing capability of the GPU, PhysX™ will provide an exponential increase in physics processing power and will take gaming to a new level delivering rich, immersive physical gaming environments
AMD HD7970
AMD CrossFire™ Technology
The AMD Radeon™ HD 7950 GPU supports AMD CrossFire™ Technology and provides excellent multi GPU scaling. With the introduction of the AMD Radeon™ HD 6900 Series Graphics, AMD began delivering improved AMD CrossFire™ Technology scaling when using two GPUs. AMD is proud to continue to offer that amazing performance with its newest AMD Radeon™ HD 7000 series graphics products.
AMD Eyefinity Technology
AMD Eyefinity Technology provides advanced multiple monitor technology enabling an incredibly immersive graphic and computing experience with innovative display capabilities, supporting massive desktop workspaces and super-high resolution gaming environments. AMD Eyefinity technology with DisplayPort connectivity enables a single GPU to support up to six independent display outputs simultaneously, depending on the OEM implementation.
AMD Eyefinity technology works with games that support non-standard aspect ratios, which is required for panning across multiple displays. To enable more than two displays, one of two cases must be satisfied:
1) Additional panels with native DisplayPort™ connectors are used;
2) Certified DisplayPort™ adapters to convert your monitor’s native input to your cards
DisplayPort™ or Mini-DisplayPort™ connector(s) are used.
With the introduction and wide availability of DisplayPort™ to single-link DVI dongles, option 2 above is a low cost way to convert a set of already-owned monitors into an AMD Eyefinity technology ready display configuration.
Unlike a similar multi monitor gaming solution offered by NVIDIA, AMD Eyefinity technology does not require two graphics cards to work. Therefore, AMD offers a multi monitor gaming solution with potential for a lower setup cost. A single AMD Radeon™ 7950 GPUs can employ AMD Eyefinity technology and AMD CrossFire™ technology to provide an outstanding multi monitor gaming experience. The following chart shows the AMD Radeon™ HD 7950 graphics in a dual AMD CrossFire™ configuration versus the NVIDIA SLI GTX 580 in three monitor gaming scenario’s.
Improved Tessellation
The latest iteration of architecture in the AMD Radeon™ HD 7970 GPU features a brand new Generation of Tesselator (Gen 9) in its Geometry Engine. This features optimizations such as; Increased vertex re-use, Off-chip buffering improvements and larger parameter caches. This helps Improve performance at all tessellation factors at up to 4x the throughput of AMD Radeon™ HD 6900 series graphics (Gen 8 ).
AMD HD3D Technology
AMD HD3D Technology is supported by an advanced and open ecosystem that, in conjunction with specific AMD hardware and software technologies, enables 3D display capabilities for many PC applications and experiences.
Your PC has evolved, offering unprecedented amount of games, photos and movies for you to play, watch, design, create, share and download in 3D. With the arrival of the latest in 3D technology you can now enjoy an enhanced visual experience on Stereo 3D-capable desktops and notebooks powered by AMD HD3D Technology.
HDMI 1.4a
More bandwith than standard HDMI allows full implementation of 60Hz Stereo 3D. Standard HDMI ports do not have adequate bandwidth for more than 3D 1080p at 24fps.
Ultra-high resolution (4k Panels)
– Current implementations currently use 4*DVI or 2*DP 1.1, which is expensive and unwieldy.
– AMD will move to a single-connector output for 4k via DP 1.2 HBR2 and HDMI 1.4a 3GHz mode.
AMD Quad Buffer SDK
– Native 3D support for Developers.
– functions with Dx9, 10 and 11
– Native support scheduled for Windows 8
Nearly 600 games support AMD’s implementation of 3D either natively or through middleware partners. Also support for many 3D displays already on the market including Samsung and LG’s proprietry implementation.
Video Codec Engine (VCE)
– H.264 HD Encoder
– Power efficient
– faster than realtime 1080p @ 60 FPS
– Variable Compression Quality
– Multiple Input Options:- (1) Frame Buffer, (2) GPU Display. (2) is useful for wireless displays.
The AMD Radeon™ HD 7950 GPU supports AMD CrossFire™ Technology and provides excellent multi GPU scaling. With the introduction of the AMD Radeon™ HD 6900 Series Graphics, AMD began delivering improved AMD CrossFire™ Technology scaling when using two GPUs. AMD is proud to continue to offer that amazing performance with its newest AMD Radeon™ HD 7000 series graphics products.
AMD Eyefinity Technology
AMD Eyefinity Technology provides advanced multiple monitor technology enabling an incredibly immersive graphic and computing experience with innovative display capabilities, supporting massive desktop workspaces and super-high resolution gaming environments. AMD Eyefinity technology with DisplayPort connectivity enables a single GPU to support up to six independent display outputs simultaneously, depending on the OEM implementation.
AMD Eyefinity technology works with games that support non-standard aspect ratios, which is required for panning across multiple displays. To enable more than two displays, one of two cases must be satisfied:
1) Additional panels with native DisplayPort™ connectors are used;
2) Certified DisplayPort™ adapters to convert your monitor’s native input to your cards
DisplayPort™ or Mini-DisplayPort™ connector(s) are used.
With the introduction and wide availability of DisplayPort™ to single-link DVI dongles, option 2 above is a low cost way to convert a set of already-owned monitors into an AMD Eyefinity technology ready display configuration.
Unlike a similar multi monitor gaming solution offered by NVIDIA, AMD Eyefinity technology does not require two graphics cards to work. Therefore, AMD offers a multi monitor gaming solution with potential for a lower setup cost. A single AMD Radeon™ 7950 GPUs can employ AMD Eyefinity technology and AMD CrossFire™ technology to provide an outstanding multi monitor gaming experience. The following chart shows the AMD Radeon™ HD 7950 graphics in a dual AMD CrossFire™ configuration versus the NVIDIA SLI GTX 580 in three monitor gaming scenario’s.
Improved Tessellation
The latest iteration of architecture in the AMD Radeon™ HD 7970 GPU features a brand new Generation of Tesselator (Gen 9) in its Geometry Engine. This features optimizations such as; Increased vertex re-use, Off-chip buffering improvements and larger parameter caches. This helps Improve performance at all tessellation factors at up to 4x the throughput of AMD Radeon™ HD 6900 series graphics (Gen 8 ).
AMD HD3D Technology
AMD HD3D Technology is supported by an advanced and open ecosystem that, in conjunction with specific AMD hardware and software technologies, enables 3D display capabilities for many PC applications and experiences.
Your PC has evolved, offering unprecedented amount of games, photos and movies for you to play, watch, design, create, share and download in 3D. With the arrival of the latest in 3D technology you can now enjoy an enhanced visual experience on Stereo 3D-capable desktops and notebooks powered by AMD HD3D Technology.
HDMI 1.4a
More bandwith than standard HDMI allows full implementation of 60Hz Stereo 3D. Standard HDMI ports do not have adequate bandwidth for more than 3D 1080p at 24fps.
Ultra-high resolution (4k Panels)
– Current implementations currently use 4*DVI or 2*DP 1.1, which is expensive and unwieldy.
– AMD will move to a single-connector output for 4k via DP 1.2 HBR2 and HDMI 1.4a 3GHz mode.
AMD Quad Buffer SDK
– Native 3D support for Developers.
– functions with Dx9, 10 and 11
– Native support scheduled for Windows 8
Nearly 600 games support AMD’s implementation of 3D either natively or through middleware partners. Also support for many 3D displays already on the market including Samsung and LG’s proprietry implementation.
Video Codec Engine (VCE)
– H.264 HD Encoder
– Power efficient
– faster than realtime 1080p @ 60 FPS
– Variable Compression Quality
– Multiple Input Options:- (1) Frame Buffer, (2) GPU Display. (2) is useful for wireless displays.
KFA GTX680
Packaging
The packaging of the KFA2 GTX680 carries on the urban theme KFA2 seem to employ with all of their products. The glossy outer sleeve hides the plain black cardboard inner box which once opened reveals a very well packaged graphics card held fast in a foam frame.
Accessories
Underneath the product sitting at the bottom of the box are the accessories which to be fair are adequate if somewhat unremarkable. More and more re-sellers are including games so it is a shame to see KFA2 have not followed suit.
Card Top
Being a reference card design we see that KFA2 have limited their artistic impression of the GTX680 to a discreet sticker on the top (and side) of the plastic GPU cooling shroud.
Card Profile
Here we see the cards used in today’s review stack atop of one another. A tower of power if ever there was one! I’m still not completely sold on the embossed, lime green GeForce GTX moniker (which cannot be removed) however when you see four cards stacked this way it is hard not to be impressed, at least to the geeks among us.
Card PCB
The black PCB is fully exposed with no backplate being used and as you can see, taking this card apart will require the removal of many screws. To the top left of the PCB you will see that NVIDIA decided to place the 5-phase VRM controller here instead of traditionally having it on the GPU side of the power circuit board.
Power Ports
The strange power port configuration serves to shorten the PCB somewhat having them on top of one another and may be easier to keep the cables tidier. However, they can be fiddly to use as the release catches on the power cables face towards one another so people with large appendages (sausage fingers) may find this configuration tricky to use.
SLI Tabs
Like all flagship NVIDIA cards, the GTX680 has twin SLI tabs meaning multiple GPU’s can be coupled together which will only be limited by your motherboard’s PCIe slots. Fortunately, the motherboard we are using today has seven PCIe slots, of which four will be populated by the graphics cards in today’s review.
Backplate
Each reference GTX680 features a HDMI port, twin DVI ports and a single DisplayPort. On top of these ports is a vented area which is relatively small compared to most GPU’s and considering this card is a closed unit, meaning this is the only area by which to exhaust how air, NVIDIA appear confident this card will run cool.
Cooling Fan
To push the air through the backplate NVIDIA use a single fan which is slightly off centre to optimise the use of the small vent at the rear of the card. In use the card is much quieter than previous GTX varients despite the fan and cooling looking very similar.
Let’s take a look inside the card shall we?
GTX680 Closer Look
As with all of our reviews we like to go a little deeper and see what makes the card effectively ‘tick’. Below is what we found with the ZOTAC GTX680 reference card but as the KFA2 GTX680 is also reference by design, we see no reason to believe this is not the same configuration.
GTX680 Cooler
The main cooler was relatively easy to remove providing you have a Torx screwdriver. Rather than using the normal Philips head screws, NVIDIA have made things a little more tricky by using Torx head which are more robust and less easier to strip than Philips but many I fear will not have a Torx screwdriver. The main core unit and I/O area has the familiar Philips head screws which when removed allowed the cooler to come away from the PCB with relative ease.
The cooling design is made up of three parts, the plastic shroud, aluminium midplate and aluminium and copper heatsink array. The shroud retains the centrifugal fan of previous designs which sucks air in directly and the forces this cool air through the aluminium finned heatsink to be exhausted at the I/O area of the card.
Heatsink
The heatsink is made up of aluminium fins soldered to an thick staggered aluminium plate. Attached to this is a machined block of copper which will make direct contact to the core. Our sample had perfect contact to the core and thus the Thermal interface material would not need to be replaced.
GTX680 Naked
As with most modern day GPU’s, the layout of the GTX680 is familiar in that the main core is framed by the (8x256MB) memory modules. Adjacent to this is the VRM module which is very different from what we are used to.
GTX680 VRM
The card operates on a 4+2 phase design. Note that there is also a redundant phase for the core which could hint at a more powerful reference card being imminent? Also worthy of note is the optional 6pin PCIe power port for additional 12v power which again could be used for future applications or perhaps it is just a throwback of an older PCB design.
Between the memory and GPU VRM is the memory controller PWM with Power sense IC’s adjacent to and above the rows of solid capacitors. These little beasties are the power limiters for the card. So where is the GPU PWM?
GTX680 GPU Power VRM
Strangely, NVIDIA have soldered this unit to the rear of the card rather than hiding it beneath the GPU cooler. This 5-phase controller hails from Richtek with a product number of RT8802A and controls the VID of the GPU.
Hynix Memory
Thankfully the GTX680 has included HYNIX memory into its armoury. With a product code of H5GQ2H24MFA R0C we know that this memory is capable of some fantastic results rated at 1.6v to run at 6Gbps so hopefully there will be some overclocking to be had there.
Kepler Core
Finally we reach the GK104 core. We have already waxed lyrical about the technological advances made so we won’t be going into detail here again. I will however say that NVIDIA are no longer implementing the integrated heatspreader seen with previous generation, instead opting for the exposed core allowing the heatsink to make direct contact with the core to aid heat dissipation.
AMD HD7970
HD7970 Front
At first glance the HD7970 looks identical to the reference HD6970. Perhaps the biggest difference in looks is that card’s cooler is gloss black rather than the matt we are accustomed to. I was hoping AMD had revised the cooling design as the 5000/6000 series were notoriously loud but looking at the cards exterior, very little has changed save for the moulded, glossy finish.
HD7970 Back
Flipping the card over we see that AMD have kept the black PCB which matches the overall looks perfectly. The cooler is attached by 12 screws and a further 4 spring load screws hold the backplate to the GPU.
HD7970 Side
The card measures 27cm long which is par for the course as far as high end cards go. The moulded finish to the card continues along the side with red plastic inserts.
HD7970 Perspective
Here you can see that rather than being the rectangular monolithic graphics card design we are familiar with, the new design is very curved yet looks strangely familiar…
Hero or Villain?
I don’t know if AMD took the cooling design from the 1966 Batmobile but the similarities are there for all to see! Joking aside, I think the new design, while hardly revolutionary looks very neat and tidy and if it performs half as good as the Batmobile it will be deemed heroic!
HD7970 Intake
The end of the card has two cut-outs for the air intake which the fan will then push over the main heatsink hidden beneath the cooling shroud.
HD7970 Backplate
The I/O area has twin mini-DisplayPorts 1.2 along with an HDMI 1.4a port and a single DVI port capable of resolutions upto 2560×1600 along with Eyefinity support.
HD7970 PCIe Power
The card uses a 6 pin and 8 pin PCIe power ports and while there is no concrete information available regarding PSU requirements, I would be extremely surprised if anything more than a 600W PSU was required. As always though, a quality branded PSU such as Corsair is recommended for optimum, reliable performance.
HD7970 Crossfire tabs
Like it’s forbear, the HD6970, the HD7970 has dual Crossfire tabs which will allow multiple GPU configurations.
HD7970 BIOS Switch
Yet another nod to the past is the Dual BIOS switch. At stock, both BIOS’s on the card are identical however, having a backup BIOS means you can update and perhaps tweak one BIOS with higher clockspeeds, voltages and fan profiles while leaving the remaining BIOS as a back-up should htings go wrong. This is a great feature to have as it will save many people from turning their expensive graphics card into an ornamental door stop.
AMD HD7970 Closer Look
HD7970 Cooler
Removing the cooler was more involved than the 6000 series due to the number of screws but with slight of hand and a raised pulse the cooler thankfully came away after gentle persuasion. The shroud comes in two parts, the plastic shell and main frame heatsink which serves to hold the vapor chamber and the fan assembly. The memory is cooled via individual thermal sticky pads while the VRM has a more traditional ‘grey’ TIM pad.
HD7970 Heatsink
The heatsink itself is a combination of a copper vapor chamber and aluminium longitudinal finned heatsink leading from front to back. From the picture above, air would be sucked in from both the vents nearest the fan end along with the fan itself. The cool air would then be passed through the heatsink and exhausted to the rear of the card.
HD7970 Fan
The fan is identical to previous incarnation and is VERY loud when running at 100%. Thankfully, the fan never got anywhere near this speed during testing and rarely budged above 40% which surprisingly was relatively quiet. Overall, the new cooling design looks very good. ‘Custom cooler’ quiet it isn’t but it’s a definite improvement, at least in terms of volume over previous versions.
HD7970 Naked
Immediately noticeable on the HD7970 is the partial integrated heatspreader. This is a new move from AMD who traditionally have never shielded their GPU’s. Unlike NVIDIA who use a full cover heatspreader to protect the delicate silicon beneath AMD have added a frame while still leaving the core exposed. Despite my own experience, I have killed a few cards by ‘nibbling’ the core with waterblocks which trust me is soul destroying so to see AMD afford some protection for clumsy folk like me is fantastic.
Surrounding the core are 12 x 256MB modules of GDDR5 from Hynix and to the right the 5+1 phase design.
HD7970 VRM
The VRM consists of Coiltronics ferrite core lo profile surface mount inductors (1007R3-R15) and solid caps which should ensure a smooth, efficient and reliable power delivery.
HD7970 ‘CHiL’ Regulator
The VRM discussed above is managed by the CHiL 8228G controller but save for AMD’s own overdrive utility allowing power control from -20% to +20% our favourite overclocking utility, MSI’s afterburner would not allow voltage adjustments at this time. I would expect this to change in later revisons though thanks to the inclusion of the CHiL controller.
HD7970 GDDR5 Memory
The memory used on the HD7970 hails from HYNIX which is great news as HYNIX memory usually grants good overclocking results. The product number of the memory IC’s is H5GQ2H24MFR R0C which is rated at 6GBps at 1.6v.
HD7970 Paste application Regulator
AMD recommended that we did not replace the paste of the HD7970 stating instead that the cooling properties would be affected and that replacement was unnecessary. However, looking at the picture above I don’t believe this particualr card had the best contact with the vapour chamber. We did however do as AMD suggest and tested the card in ‘stock’ form. Replacing the TIM with MX-4 saw a load temp decrease of 2c idle and 5c load so despite AMD’s claims, I may be tempted to replace the paste should you feel confident and competent in doing so.
HD7970 ‘Tahiti’ Core
Finally we reach the beating heart of the AMD HD7970, the GPU core codenamed Tahiti. Unlike older AMD/ATI cores, there were no markings whatsoever on the silicon itself with the product information instead being found on the IHS frame.
Test Setup
With the latest Intel chipset release we decided to update our ageing X58 setup to the new X79 chipset. Not only that but we will be using the flagship Core i7-3960X Extreme Edition processor clocked to 4.5GHz to ensure we have no CPU bottlenecks. With 16GB of ram on board there should be minimal paging so the benchmarks we run today will hopefully give a true reflection of the graphics card’s performance.
Some of the components for our test bench were kindly donated by CORSAIR who provided the majority of components without whom we would not have been able to bring you this review so a big thanks to them.
Graphics cards on test:
KFA2 GTX680 2048MB GDDR5
AMD HD7970 3072MB GDDR5
For this review we will be using the previous refresh of NVIDIA drivers due to the decision NVIDIA made in disabling PCIe Gen 3 in their latest driver set for X79. While there are advantages to both driver sets, we are comparing like for like today so as AMD drivers are PCIe 3.0 compatible, we didn’t want to confuse matters by using PCIe 2.0 with the NVIDIA setup. The AMD drivers used today will be the CAT 12.3 drivers.
Motherboard: MSI X79 Big Bang X-Power II
CPU: Intel Core i7-3960X Extreme Edition @ 4.5GHz
RAM – 16GB (4x4GB) Corsair Vengeance (Red) DDR3 @ 1866MHz 9-10-9-27
Power Supply – ANTEC High Current Pro 1200W
Hard Drive – Hitachi SATA II 250GB
Samsung F3 1TB
Cooler – Corsair H100
Benchmarks
We have also updated our range of benchmarks to include the following games which will give the graphics cards a challenge and sort the wheat from the chaff. Each benchmark will be run at both stock and maximum overclock settings and utilise 1920×1080 and 2560×1600 resolutions along with AA at various levels (application dependent):
Battlefield 3
Crysis 2
Crysis Warhead
Metro 2033
Batman: Arkham City
Elder Scrolls: Skyrim
DiRT 3
Futuremark 3DMark 2011
Unigine: Heaven 2.5
Crysis 2
Crysis Warhead
Metro 2033
Batman: Arkham City
Elder Scrolls: Skyrim
DiRT 3
Futuremark 3DMark 2011
Unigine: Heaven 2.5
Let’s see how today’s cards performed…
Overclocking, Temperature & Power Consumption Analysis
Overclocking
As you can see from the picture above we ran the cards using the older Forceware drivers which have PCIe Gen 3 enabled.
For an overclocking comparison we will compare 1 card versus another. There are so many configurations we could use when comparing multiple setups so in this section we will just compare mano-a-mano, 1 vs. 1.
Overclocking the GTX680 will primarily feel a little odd as you don’t directly overclock the frequencies, rather you set offsets/limits the card can works toward depending on the power target you also set. In effect you are setting the overclocking limits of the card, not what is actually achieved as the GPU will decide exactly how much boost is required and adjust the clockspeed accordingly. This is is what NVIDIA call GPU boost.
We managed to hit 1240MHz on the KFA2 card with the power limit set to 132% which is a good achievement considering the stock speed is already breaches the 1GHz barrier at 1006MHz. As expected, the memory also reacted very well to overclocking too reaching 1550MHz (6200MHz effective).
Our OEM AMD card was also a tidy overclocker hitting the milti with AMD Overdrive at 1125MHz from a stock clocked (above) 925MHz. Not satisfied with this we went a step further with the Sapphire TRIXX utility which allowed us to clock the HD7970 to 1255MHz on the core and an amazing 1720MHz (6880MHz effective) on the memory with the voltage slider set to maximum.
For giggles, we ran both 4x GPU setups in their max overclocked state but our poor 1200W Antec High Current Pro model didn’t appreciate the 1340W load we placed upon it(!) and decided to take it’s ball home. Thankfully, it was simply the load protection of the PSU kicking in and so no serious damage was done, other than a toasted windows install and my own cardiac arrest so if you do plan on running four of these cards in a heavily overclocked state I would recommend a 1500W PSU as an absolute minimum. Thankfully, as you will see below, there were no such problems running the setups at their stock clocked levels which is how we will be comparing the configurations today.
Power Consumption
To test for power consumption we do not use anything too technical, just a wall meter which measures the total system power draw minus the monitor. For load tests we use FurMark (mutli-GPU) which fully loads all of the GPU’s. We then take a measurement when the temperatures of the cards have levelled out to ensure an accurate (where possible) power consumption reading. While Furmark stresses graphics cards to their limit, normal gaming scenarios do not add the same level of stress so the figures below are a ‘worst case’ scenario.
As you can see from the chart above, in single card configuration, it is pretty much even however when multiple cards are added, the NVIDIA setup becomes the less hungry, a trend which continues in 2x, 3x and 4x configurations.
Temperatures
To test temperatures we measured idle temperatures after booting windows, letting all applications finish loading and ran a few benchmarks. Once the benchmarks were complete we left the card to reach a cooling plateau where we then took the idle temperatures. For the load tests, we ran Furmark for 20 minutes, taking the absolute maximum temperature attained.
Another victory was had by NVIDIA with the card(s) running quite a lot cooler than their AMD counterparts. The difference in cards can be summarised simply by one AMD HD7970 hitting the same temperature as four GTX680’s together. Again, in ‘normal’ gaming circumstances, the temperatures will likely be lower than this. One point to note however is that the NVIDIA cards clock frequencies began to throttle back slightly when temperatures exceeded 70c which no doubt assisted NVIDIA in controlling the temperature of the GPU. NVIDIA would call this a ‘feature’ of GPU boost and indeed it is good that the card has some form of thermal protection rather than simply crashing. Comparing the two cards temperatures however becomes difficult because on testing I did not notice any such throttling of the AMD cards.
Let’s move on to what we all want to see – just how well the different setups perform and compare in our suite of benchmarks…
Battlefield 3
Battlefield 3 leaps ahead of its time with the power of Frostbite™ 2, the next instalment of DICE’s cutting-edge game engine. This state-of-the-art technology is the foundation on which Battlefield 3 is built, delivering enhanced visual quality, a grand sense of scale, massive destruction, dynamic audio and incredibly lifelike character animations. As bullets whiz by, walls crumble, and explosions throw you to the ground, the battlefield feels more alive and interactive than ever before. In Battlefield 3, players step into the role of the elite U.S. Marines where they will experience heart-pounding missions across diverse locations including Paris, Tehran and New York.
Average and minimum frames per second were measured using Fraps during the parking lot scene on Operation Sword Breaker. High and Ultra presets were loaded at both resolutions.
Crysis 2
Crytek took an internet beating from its loyal fans due to the fact that they pandered to the console masses and dumbed down the graphics to DirectX 9. Fortunately, fans of the PC version of Crysis managed to force Cryteks hand and they have since developed a DX11 patch and have also enhanced the textures with added tesselation. It is for this reason that we have decided to include it in our suite of benchmarks.
To test the cards we used the Crysis 2 benchmark courtesy of Adrenaline with both the 1.9 patch, DX11 update and Hi-res texture pack. Settings were manually placed to Ultra, DX11 and tessellation enabled with the Central Park map being run 3 times per resolution benchmark test.
Crysis Warhead
Following the success of Crysis, the stand alone expansion “Warhead” was hotly anticipated to bring more game play, more weapons, improved visuals whilst maintaining better performance, the latter being what gamers wanted most. After all how many people upgraded their rigs just to play this game… In the game, you play as the upbeat Brit Sergeant “Psycho” Sykes, chasing down a Korean nuclear warhead. Throw in some ice shooting aliens and we’re bound to have a great game. It sure delivered.
Testing was done using Framebuffer’s Crysis Warhead benchmarking tool. The tests were configured to DirectX 10 at Enthusiast levels. Each test on the Ambush map was looped three times from which the average FPS was recorded.
Batman: Arkham City
Batman: Arkham City builds upon the intense, atmospheric foundation of Batman: Arkham Asylum, sending players soaring into Arkham City, the new maximum security “home” for all of Gotham City’s thugs, gangsters and insane criminal masterminds. Set inside the heavily fortified walls of a sprawling district in the heart of Gotham City, this highly anticipated sequel introduces a brand-new story that draws together a new all-star cast of classic characters and murderous villains from the Batman universe, as well as a vast range of new and enhanced gameplay features to deliver the ultimate experience as the Dark Knight.
The results below were obtained from the builtin benchmark utility from the game. All the settings were set to
The Elder Scrolls V: Skyrim
Skyrim is not a direct successor to the very popular Oblivion. Instead it takes place 200 years after the events of Oblivion. The non-linear playing that made the Elder Scrolls such a popular series does however make a return along with the beautiful scenery the encompasses the world of Skyrim.
Average and minimum frames per second were measured using Fraps during the drop down passage to Riverwood. V-sync was disabled to allow fluctuating (higher the 60) FPS for these runs. It is however recommended that v-sync is left enabled during normal gameplay due to the games dynamics.
DiRT 3
As players race to elevate their global standing, DiRT 3 is back delivering mud, sweat and gears world over: from the intense weather-beaten rally stages of Europe, Africa and the US, to executing performance driving showcases and career challenges where car control is pushed to spectacular limits.
The in-game benchmarking tool was used to test performance. All the display settings were set to the maximum available and DirectX 11 was applied. The average FPS was recorded at the end.
Metro 2033
Based on the novel by the same name, Metro 2033 is a first person shooter with elements of survival horror. As Artyom, you trawl through the mutant infested metro systems of post-apocalyptic Moscow. The game delivers presents a grim ambience that reflects on the overall atmosphere of the underground setting. Metro 2033 uses the setting as a back drop to showcase some of the best visuals of any games out now, making it the ideal benchmark for those looking for the best graphics card.
The bundled benchmarking tool was used on Metro 2033, configuring the settings to High and making use of DirectX 11. The frontline scene was looped 3 times for each configuration. The average FPS was recorded at the end.
Futuremark: 3DMark 2011
3DMark 11 is the latest version of the world’s most popular benchmark. Designed to measure your PC’s gaming performance 3DMark 11 makes extensive use of all the new features in DirectX 11 including tessellation, compute shaders and multi-threading. The benchmark was run in both Performance and Extreme presets.
Unigine: Heaven 2.5
Heaven Benchmark is a DirectX 11 GPU benchmark based on advanced Unigine™ engine from Unigine Corp. It reveals the enchanting magic of floating islands with a tiny village hidden in the cloudy skies. Interactive mode provides emerging experience of exploring the intricate world of steampunk.
Graphics Card Scaling – Gaming
Perhaps the most important aspect to consider when comparing multiple graphics card configurations is the performance scaling – how much performance extra graphics cards will give. It’s all good and well having a wall of GPU’s in your system but if they nothing to the performance then there really is no point other than to show off just how much money you have to waste (and how little you care about the environment!).The graphs below are an amalgamation of the various settings used in each game/benchmark to give you an overview of how much performance adding an extra card will give you when compared to lesser configurations. I will however concede that their are an infinite amount of settings and configurations that may give different results. For example the higher the resolution, the higher GPU scaling is likely to be as that added GPU grunt will be put to use. So with that said, the results below are only intended to give an indication of performance gains based on the overall performance of the settings used in our run of benchmarks…
Gaming Benchmarks
Battlefield 3
Dual AMD cards tend to scale better in EA’s shooter with the FPS almost doubling compared to a 68% increase given by NVIDIA. Tri-SLI gives the least increase in FPS giving less than half the performance increase of AMD while in 4x configuration, both setups offer less than a 20% improvement over triple card setups.
Crysis 2
Crysis 2 favours SLI over Crossfire but clearly anything more than a dual card configuration offers nothing. Indeed in our testing the performance of the test rig actually suffered with anything more than a dual card configuration!
Crysis Warhead
The tables were turned once more when we ran the older Crysis Warhead game with Crossfire once again scaling very well, offering over 80% increase in FPS. Once more however we see that four cards did not offer any increase in FPS (Quad SLI impeded performance).
Batman: Arkham City
Out of all the games tested today, Batman offered the least in terms of multi GPU comparability. While there were minimal gains in the varying SLI setups, anything more than dual AMD configurations resulted in a drop in frames per second.
The Elder Scrolls: Skyrim
Skyrim favours SLI over Crossfire here yet once again we saw a slight performance deficit when SLI was used in 3 and four card configurations. AMD too showed a performance decrease when four cards were used with 3 cards showing only a minute gain in performance.
DiRT 3
Now this is more like it. All of the graphics card setups, be it NVIDIA or AMD showed performance gains were to be had, albeit with diminishing returns the more cards you add. A 92% and 54% increase by AMD and NVIDIA respectively is nothing to be snuffed at and once again we see AMD have the advantage when mutliple GPU’s are added.
Metro 2033
This time we see NVIDIA hold the scaling advantage with all 3 different setups scaling better than their AMD adversaries. All the configurations did however showed performance increases were to be had thanks to the high graphical demands of Metro 2033.
Let’s move on to the benchmarking area where those most likely to have such a setup will reside…
Graphics Card Scaling – Synthetic Benchmarks
Futuremark: 3DMark 2011
Above you can see just how much more performance scaling increases the harder the cards are pushed. It is clear that in many of the gaming results, the multi card setups were not being pushed enough and that they demand intensive, ultra high settings and resolutions to stretch their proverbial legs. In the case above, we see that AMD once again scale better than NVIDIA in both Performance and Extreme scenarios.
Unigine: Heaven 2.5
The Heaven benchmark shows the biggest improvements of all of the benchmarks used today. The final benchmark of today’s review cements AMD’s place as the no#1 GPU setup, at least in terms of scaling performance, bettering the green team by a fair margin.
Let’s move on to the conclusion where I try to put all of the results we have seen today into some form of cohesive verdict…
Conclusion
Whichever your preference of graphics card manufacturer, whether you sit with the green team or the red, both factions will serve you extremely well. If I were forced to choose between the two as a single graphics card, the KFA2 GTX680 wins out. It wins in all areas: quieter, cooler, faster while consuming less power. For that it deserves considerable praise.However, dual card setups and higher have to go with the better scaling AMD CrossfireX setup. The NVIDIA configurations simply could not compete with AMD in the scaling of single vs multiple card setup. While it could be argued that I was using an older driver (PCIe 3.0 enabled) for the NVIDIA setup, a quick and dirty run using the latest drivers showed comparative results. Any more than a dual card configuration and it becomes less clear as it would depend entirely on your setup and choice of game or benchmark.
Some games favour AMD while in others, NVIDIA have the clear advantage. What is obvious though is even a 30″ 2560×1600 screen does not warrant a triple graphics card setup, be it the HD7970 or GTX680. The performance gains offered at this resolution, even with Ultra settings applied are so small that you would be very hard pressed to physically notice the difference other than for benchmarking purposes. The diminishing returns only worsen when a fourth card is added. To make matters worse performance was often lesse with four cards than three! I’m sure future driver revisions would improve performance but I cannot stress more clearly that unless you intend of running Eyefinity or NVIDIA Surround with multiple, large screens then you would simply be wasting your money with a quad setup from either camp.
So while the AMD cards scale better and the NVIDIA card is the current single GPU performance king, it is NVIDIA who have the quiter and less power hungry card. In quad configuration both setups consume a serious amount of energy. You will need nothing less than a 1200W PSU for a quad setup with a 1500W recommended as a minimum should you decide to push the cards as far as they will go. That is going to make an almighty dent in anyone’s electricity consumption but if you can afford to buy four of these flagship graphics cards, it is unlikely you are going to worry about the electricity bill. Four AMD cards are cheaper to buy, costing around £1500 compared to NVIDIA’s £1776 and at the settings used today appear to work better when paired together.
The story doesn’t end there though as the NVIDIA setup ran significantly quieter than the AMD setup. This could be because of throttling issues due to excessive temps ensuring NVIDIA’s cards temperatures and therefore fan speed was kept artificially low. To counter this I set both setups fans to their maximum and both setups were extremely loud at full tilt. It has to be said though that the NVIDIA cooler was quieter than AMD albeit slightly and with four cards sandwiched together, whichever setup you choose, noise will be an issue you will have to endure I’m afraid.
If you happen to have around £2k laying around to buy these mammoth GPU setups with change leftover for PSU to run them, you will certainly have the bragging rights to put anyone in their place. If however you do not have large multiple screens to run and a heavily overclocked CPU to feed them then I fear many will be sniggering behind you back because what four graphics cards can do, the smart money would be spent on a Crossfire/SLI setup which uses just two GPU’s and as we have shown today often performs better.
With regards to today’s samples; the KFA2 cards are very good, clock equal to other GTX680’s we have tested and come presented in a well constructed, if somewhat uninspiring package. Price wise they are very competitive against other reference GTX680 cards and thus are likely to attract attention. If you are in the market for a GTX680 then you could do a lot worse than the KFA2. For these reasons the KFA2 card is granted our Silver award. Not the best GTX680 out there but it is worthy as a reference card and beats the AMD equivelent in terms of performance, power consumption and noise.
I was extremely tempted to give the GTX680 SLI setup our Elite award, simply because they are ridiculously fast. I will however refrain from doing so because while 2 and 3 cards do perform extremely well, they are out stripped in raw multi GPU frames per second performance by their AMD equivalents, the HD7970. Moreover, both multi GPU configurations, be it 3 or 4 card AMD or NVIDIA setup, simply does not offer any worthwhile improvements unless you are the most hardcore enthusiast who craves every last point from a benchmark or have multiple screens with which to fully utilise the raw GPU processing power.
Pros:
– Silly fast FPS figures
– Blistering Performance
– Cool running
– Low Noise
– Good overall package
– Small dimensions
– Competitive pricing
Cons:
– Do not scale as well as AMD cards
– More expensive than the AMD competition
– Silly fast FPS figures
– Blistering Performance
– Cool running
– Low Noise
– Good overall package
– Small dimensions
– Competitive pricing
Cons:
– Do not scale as well as AMD cards
– More expensive than the AMD competition
Click here for an explanation of our awards at Vortez.net. Thanks to Overclockers UK for providing today’s review samples. Buy the KFA2 GTX680 2GB GDDR5 HERE.
