These things can take time, but at last this generation of graphics cards has been pretty much finalised. AMD did its part early on, having almost its entire lineup sorted before the summer, while Nvidia has been dragging its heels like a schoolboy who's forgotten his homework.
The release of the GTX 650 Ti, though, marks the completion of this generation of cards, and despite the fact it's taken nearly an entire year to get to this stage, things have never looked so good for the consumer.
It's somewhat inevitable that every generation offers more power for less cash than ever before, but it's still surprising that a card you can buy for less than $150 will be able to deliver more than 30fps on average at full HD resolutions in the latest DX11 titles.
It's strange, then, that people are calling these the end times for the discrete graphics card, what with all the advances in integrated CPU graphics performance. For the end user though, us PC gamers, our rigs have never looked so sprightly.
Being able to pick up mid-range cards with the sort of performance the high-end cards of the past few years could only wish for, is making our gaming experience the envy of the slavering console hordes.
Just look at the gaming shows of the past year, GamesCom and E3 especially. All the best-looking titles were being demoed on high-end PCs. Why? Because there's no way to get the games looking as good on the current generation of consoles.
With the next generation of those limited tech bricks not arriving until the tail-end of 2013, the PC has an entire year of even greater gaming dominance. And by then, chances are it's still going to hold the technological gaming advantage.
But what of right now? Who are the winners and losers in the great graphics arms race? We've lined up the full range of gaming-capable graphics cards of this generation and we'll tell you where the smart upgrade money should go.
2012 was a rather strange year in terms of graphics card releases. It all kicked off in a rather bizarre way just before Christmas 2011 when AMD 'launched' its HD 7970 - despite only actually having stock in January.
We could have understood AMD's desire to get the card in front of the tech press before Christmas if Nvidia was likely to have its brand new card out soon after, but the green team was a long way behind with its new GPUs - but we'll get back to that later.
Still, the new top-end card arrived and it was - somewhat inevitably given the lack of Nvidia competition - the fastest GPU around. But it was a brand new GPU architecture, in a year when AMD had also released a new CPU architecture with limited success.
That new architecture represented a strange flip in terms of the focus of each side's graphics card design. Rather than looking to continue with its standard, one-trick method of hitting gaming frame rates, AMD was now, with the Graphics Core Next chips, focusing on nailing compute performance. That had been Nvidia's route with the previous generation Fermi cards and now AMD was creating a very similar architecture to cope with the graphics engines of the day.
And then in March, Nvidia dropped the K bomb: Kepler hit the streets with the GTX 680. It's new 28nm architecture had almost switched sides with AMD - the Kepler GPUs focused more on raw graphics performance than the general compute-focused Fermi cards that came before it.
Where's my compute?
Nvidia has left a lot of the compute power out of the consumer-focused cards then, preferring instead to jam the extra silicon necessary for such general-purpose calculations into the Tesla series of professional cards.
That space-saving silicon design means the top consumer Kepler chip is an incredibly power-efficient graphics processor, with a TDP of just 195W compared to the 250W TDP of the competing Radeon HD 7970. Nvidia's Kepler GPU also introduced the Intel-aping GPU Boost, which dynamically ramps the GPU clock speed depending on the power usage and thermal envelope.
Now it was AMD's turn to play catch-up as it introduced the PowerTune Technology with Boost - which more or less did the same thing for its GPUs - later in the summer. In terms of raw graphics performance, the top two tiers of graphics cards - GTX 670/680 and HD 7950/HD 7970 - are much of a muchness, trading performance leads in different game benchmarks.
At the top, then, it pretty much comes down to which card's feature-set you prefer and which plays best with your favourite games. Things have become a lot less equitable in the lower echelons of the graphics card market. While AMD had released its full HD 7900, HD 7800 and HD 7700 series of cards into the wild by the end of March, Nvidia would take another seven months to actually flesh out its Kepler lineup.
That meant AMD could keep its prices high while it was the only new kid on the block, and then take an axe to its GPUs' price tags when Nvidia's rival cards finally touched down. We will probably never know the real reason for Nvidia's tardiness on its lower-end cards, but rumour has it that low silicon yields on its 28nm process had something to do with it.
Nvidia just claimed to be waiting for the wealth of big games titles coming out at the end of the year. Whatever the truth of it, AMD is winning the price battle, and has better cards available at every price-point in the sub-$400 market.
One of the big problems, though, is the fact there are few games out there at the moment that will really tax a modern GPU. If you've bought a card in the past couple of years and are still sitting on a native resolution of 1,920 x 1,080, or lower, then there's really not a compelling reason to upgrade. Chances are you'll still be hitting reasonable frame rates in the latest games and that means the upgrade market is a little flat.
If you are looking for a new graphics card, though, there has never been a better time to make the move because you can pick up an inordinate amount of pixel-pushing power for very little cash these days. But are you still going to need a graphics card in your next PC?
Both AMD and Intel are making great strides in the integrated graphics market with future generations of chips. AMD, though, is segmenting things, leaving its top consumer chips, the FX range without any graphics on die. Intel, on the other hand, is resolutely sticking its iGPUs in every chip it throws out, bar the crazy £800+ Sandy Bridge E side.
Discrete days numbered
With the processor upgrade to Haswell next year, Intel is predicting a doubling of its graphics performance in Ivy Bridge. In terms of gaming on the desktop that still isn't going to give you the frames per second you'll need to get the most out of Star Citizen, but on the mobile side it's talking up the capabilities of the GT3E Haswell graphics.
Speaking to the graphics guys at IDF this year, Intel seems to think it will be making dedicated gaming laptops without dedicated graphics cards a reality in the next generation. With the next generation of AMD's APUs looking to include the proper Graphics Core Next architecture as the GPU component, its new lineup could also make things sticky for the dedicated graphics side.
With no concrete news yet of Nvidia's Project Denver - it's own ARM-based gaming CPU with a full-fat GeForce graphics part - the green side of the graphics war is looking a little ill-equipped for the coming battle - at least on the PC.
The next generation of both AMD and Nvidia graphics cards will simply be building on the advances of their respective first 28nm parts, so we don't expect a lot of particularly interesting new tech next year, just a little faster and a little more efficient. So that makes this generation even more interesting.
These cards will consistently drop in price up to the launch of the slight upgrades of the HD 8000 and GTX 700 series of cards, and with the next generation of consoles not arriving until the tail end of next year, right now this is all the card most of us need. But which one is best for you? Let me show you…
No comments