So here it is, the AMD Radeon HD 7970 and, for the time being, it’s the fastest graphics card around.
AMD has blinked first and opted to release its brand new graphics card architecture before Nvidia, and just before the new year.
It’s a brave move by AMD though. Bringing out a radically different graphics design spec, compared with its previous vector processors, in the same year as it brought us a brand new CPU architecture.
Especially given the depressing failure of the AMD FX chips.
We’ve seen little bits from AMD about its new architecture, the plainly-named Graphics Core Next, before, but now it’s humming away in our test bench and throwing pixels and polygons around our high-res screens without a care in the world.
Well, maybe the AMD Radeon HD 7970 does have some cause for concern considering Nvidia is set to launch its newest graphics architecture, code-named Kepler, in the early Spring of 2012.
Maybe that’s why the timing of this release is so odd.
AMD has decided to effectively launch the Radeon HD 7970 just three days before Christmas, a notorious deadspot in technology news. We’ll see the ‘true’ launch of the card, with the manufacturer’s versions, coming early in the new year.
Though speaking with AMD board partners they’re not allowed to release overclocked cards until well after launch.
So all they will be releasing are these reference Radeon HD 7970 cards with new stickers.
That means much of their thunder will be stolen by this here unveiling. And even that thunder will be deadened by the deafening silence of the holiday period.
So, like a junior cabinet minister in a time of crisis, is AMD trying to bury the launch of the Radeon HD 7970?
Maybe it realises the market for a £450 graphics cards is absolutely minute. Maybe the forthcoming Radeon HD 7950 is the card that it wants to really concentrate on. Maybe it just wanted to make sure us tech journos had to work right up to Christmas this year.
Whatever the real reason for this staggered launch is, it’s time we took a proper look at what makes this here AMD Radeon HD 7970 tick.
In terms of the actual make up of the AMD Radeon HD 7970 it’s a fairly different beast to the previous, Cayman GPU-powered AMD Radeon HD 6970.
This card represents a new direction in GPU tech for AMD as well as a brand new production process and GPU technologies.
In terms of firsts AMD has them all pretty much nailed with the Radeon HD 7970. It’s the first GPU to be built with miniscule 28nm transistors, the first DirectX 11.1 graphics card and the first official PCI Express 3.0 component too.
Of all those the really interesting one is the 28nm die shrink that comes with this new Tahiti core. The Tahiti GPU is going to be the chip powering the top end cards and that in turn is built on the new Graphics Core Next architecture.
The whole HD 7000 series also goes by the name of Southern Islands, hence the Tahiti tag.
That die-shrink means that it can jam more components into a smaller chip footprint and that should mean more power to you.
And compared to the number of 40nm transistors in the AMD Radeon HD 6970, a paltry 2.6 billion, it has done a good job of squeezing more into the Radeon HD 7970.
This card has 4.3 billion transistors in the GPU itself.
That there is a frightening number, and AMD has a lot more of those up its sleeve for the Radeon HD 7970.
AMD’s claimed compute performance for the Tahiti XT GPU inside the HD 7970 is one such number. At 3.79 TFLOPs there is some impressive number crunching capabilities buried inside this chip.
All it needs is the software to take advantage of it.
The compute performance has been the driving factor in the change from Northern Islands to Southern Islands, and from the traditional vector style processor to the scalar processor that makes up this new GPU.
Previously AMD had focused most of its silicon towards the goal of making a card specifically for graphical processing, leaving the more ephemeral notion of general purpose GPU computing to Nvidia and its CUDA cores.
So it put all its eggs in the four-way vector processor architecture basket. Essentially that meant sorting out single instructions into batches before firing them down the GPU pipelines. It was a much more elegant solution and for doing fixed graphical processing it was incredibly efficient.
The resulting HD 4xxx thru HD 6xxx series cards were great pixel pushers at impressively lower power requirements than their peers.
Now though AMD wants a slice of the GPGPU pie and as such has shifted to a similar scalar architecture to that used by long-time rivals Nvidia.
That means it’s going for a more brute force approach which involves having a large number of simple processors in an array, giving them each one thing to work on at a time until all the instructions have been completed.
The Graphics Core Next architecture then is built of Compute Units, which are similar in nature to teh Streaming Microprocessors Nvidia introduced with the Fermi architecture.
The Tahiti XT GPU in the Radeon HD 7970 is built up of 32 of these Compute Unites which are essentially self-contained processors capable of acting independently of the whole.
Inside each of the Compute Units are four vector units, and in each of those is 16 unified shaders or stream processors or Radeon Cores. Depending on who you talk to.
That means in the full Tahiti XT you’ll find 2,048 shaders, and compared to the 1,536 shaders of the HD 6970 that’s a fair mark up.
It’s this combination of the four-way vector processing and the scalar architecture that AMD hopes will push its Tahiti-based GPUs to the top of the graphics card pile.
Though it’s not just the number of Compute Units that makes the difference in performance terms, the actual core clock of the Radeon HD 7970 is far higher than previous cards too.
At 925MHz out of the box it’s a chunk faster than the 880MHz of the Radeon HD 6970.
So, what do all these new architectural shenanigans mean for the graphics card’s performance then? Let’s find out…
We’ve put the AMD Radeon HD 7970 through its paces in a raft of different benchmarks to see how it stands up against the competition.
The competition comes in the shape of the previous fastest single-GPU card available, the Nvidia GTX 580, and the previous generation of AMD’s top single-GPU card, the Radeon HD 6970.
As these are the top of their respective lines we’ve pushed them to the max running all our benchmarks at the ultra-high, 30-inch panel resolution of 2560×1600 and all with 4x anti-aliasing too.
The results show a definite gaming win for the new AMD card, with the HD 7970 coming out on top in every single one of our gaming tests.
The two compute tests we used though don’t really show up the enhanced computing power of the new architecture, though that’s maybe more to do with the software not utilising the new silicon effectively.
Still, the GTX 580 seriously outperforms the new AMD card in the HD H.264 GPU encoding benchmark, though the HD 7970 has it beat in the Open CL-based GPU Caps Viewer test.
DirectX 11 tessellation performance
DirectX 11 gaming performance
DirectX 10 gaming performance
Open CL performance
Platform power performance
Now we come to the crunch then, the performance of this brand new slice of graphical silicon goodness.
Well, pretty quick.
The AMD Radeon HD 7970 has got the Nvidia GTX 580 beat hands down when it comes to raw graphics chomping, pixel-pushing, polygon-smashing performance.
In all of our gaming benchmarks the HD 7970 has the Nvidia card in second place, especially when you look at the particularly demanding Heaven 2.5 and Crysis 2 benchmarks.
With both those benchmarks you’re looking at around a 30% improvement over the Nvidia GTX 580.
Those are the best case scenarios though as the performance benefits in the other titles are much less pronounced, sitting between 13% and 20% frame rate improvements.
When it comes to the previous generation, the venerable Cayman-powered HD 6970, things look better still. The HD 7970 posts a 65% improvement over the older card in the tessellation-heavy Heaven 2.5 benchmark.
That’s the highest performance increase, but at worst it’s around 25% faster. Ideally you’d be hoping for at least half again performance from a new generation, especially a die-shrink, but that can happen if you take things into your own hands.
This is all before you start waving around the overclocking stick then.
And boy, does this chip overclock.
In fact we ran into the limits of the AMD Overdrive software rather than the limits of the hardware itself. We pushed the card all the way up to 1,125MHz on the core clock and 1,575MHz on the memory clock.
With those numbers the performance increase over the GTX 580 and HD 6970 are way more pronounced with the HD 7970 beating the previous generation by up to 80%.
When overclocked too it gives the previous generation of dual-GPU cards, the Nvidia GTX 590 and HD 6990, a good run for their money. The GTX 590 had the highest score of 33FPS while the HD 7970 comes in just behind at 32FPS.
The HD 6990 actually lags behind it with 29FPS.
It’s not just all about the raw performance figures though as AMD has made a lot of effort with the power requirements of the HD 7970, especially when running in idle mode, and with the screen off.
The ZeroCore Power technology means when the screen turns off and the machine goes into the ‘long idle’ state – where the PC is still running but there is nothing being updated on the screen so the panel goes to sleep – there is only a single chip on the card still running.
That chip is there just to tell the PC there is still a card in the PCIe slot and not to worry. The rest of the graphics card turns off completely, even the fan.
We’ve accounted for the architecture and performance of the AMD Radeon HD 7970, but there is one thing we haven’t covered, the price.
And that’s key to the graphics card battle and in the case of the HD 7970 may well be the thing that truly buries it.
At approximately £450 it’s quite frankly a ridiculously priced card.
It may well be the fastest single-GPU card around at the moment, but there is still little justification for the price.
Apart from the fact that a chip with 4.3 billion transistors, running on a new production process, is going to be rather expensive to manufacture that is.
There’s also the question of whether you really need the levels of graphical performance the HD 7970 offers. There are very few of us out there running a monitor capable of the eye-watering resolutions of 2560×1600 so realistically a 1920×1080 resolution is going to be more likely.
And at that resolution the excellent £365 Nvidia GTX 580 is all the card you’re going to need.
Even if we were going to give AMD the benefit of the doubt, and trust that it wont mess around with pricing across the Atlantic, taking the US price of $549 and the current exchange rate, plus VAT you’re still looking at around £430.
That’s still too much for a new graphics card these days.
How much is the forthcoming dual-GPU version going to cost? £700?
It’s a shame, as if the card had come out at the same sort of price as AMD originally tagged the HD 6970 with it would have had a far better reception.
There are some good points about the HD 7970 the pricing cannot diminish however and that’s because they will be rolled out across the Tahiti line.
That’s the impressive ZeroCore Power tech.
Being able to shut the GPU down almost completely when idle is a great feat of engineering, made doubly so when you bring in CrossFire setups.
When in CrossFire mode all the GPUs will be in use when you’re gaming, but when you drop down to 2D desktop mode all GPUs, bar the main card attached to the display, will shut down, fan and all.
That’s impressive and means when you’re using your monster rig in general computing tasks you’re not going to require your own private Sellafield to power it.
The overclocking potential of the HD 7970 is likewise impressive.
We don’t know the limits you can push this card to however as the software topped out before the hardware did.
Still we managed a huge overclock which makes us question why it wasn’t rated as the first 1GHz GPU out of the box.
That though was answered by AMD’s Director of Product Management for Discrete Graphics, Zvika Greenstein, at a recent tech insight event for the Graphics Core Next products.
“One of the things the enthusiast likes to do with our cards is overclock it, they pay a premium for that,” says Greenstein. “We can position the HD 7970 as the fastest graphics card in the market at the reference clocks so we thought that we might as well leave it to the end users.”
In essence, AMD didn’t need to push the silicon, it’s going to rely on the actual board manufacturers to do that themselves.
And they’ll in turn charge end-user a premium for it.
We also had a few driver problems with the card too. We couldn’t get any reasonable performance numbers out of the HD 7970 in our new Sandy Bridge E machine, with the benchmarks falling way behind the GTX 580.
It was only when we switched to our second AMD FX-8150 powered setup (the first blew up on the second benchmark) that we started to see proper performance numbers.
So in the end it’s a tough ask for us to recommend going out and picking up the AMD Radeon HD 7970.
Especially when we know the Nvidia riposte is only a few scant months away. And according to Nvidia insiders it’s quietly confident about its chances of the top Kepler card besting AMD’s Tahiti XT-powered HD 7970.
Again then it’s a wait and see game.
If Nvidia’s card is even more expensive than the HD 7970 then this card may start to look like good value.
More likely the second tier Tahiti-powered Radeon will be the card that we really want to recommend.
Especially if it comes with the good parts of the Tahiti core, namely the ZeroCore Power tech and the heavenly overclocking headroom, all at a reasonable price.
The overclocking potential of the AMD Radeon HD 7970 is incredible.
Topping 1,100MHz is a huge overclock and makes it almost comparable to the previous generation of dual-GPU cards.
With or without an overclock though it is most definitely the fastest single–GPU graphics card around.
Unfortunately the performance boost the HD 7970 offers isn’t enough to really justify the vast price tag AMD has lumbered the new card with.
At this price it surely isn’t going to sell in any volume.
The shadow of driver problems still loom over any new AMD graphics release too.
It may well be the fastest single-GPU card around, but the price is absolutely prohibitive. At £350 it would have been a hit, as it is we have no choice but to look elsewhere for a GPU recommendation.
Post from Techradar