There has been some pretty grim reading surrounding AMD in the last few days. The fact it's now worth just a quarter of what it paid for ATI back in 2006 may sound pretty devastating, but even just a year later in 2007, the company was actually worth less than the $5.4 billion it paid for the GPU giant
, so things have clearly been on a consistent downward spiral.[break]
That said, it has seen a few rebounds, namely in 2013 when it was tied into new Playstation and XBox launches, but despite a number of competitive GPUs coming from AMD in the last few years (the Radeon 5000, 6000, 7000 and R9 200-series ranges all had stand out cards that were able to do battle with Nvidia, either on price or performance or both), the company has yet to maintain an extensive upward streak for more than a couple of months since the aquisition of ATI and the launch of Phenom.
Its CPUs are clearly a huge part of the reason for this. They just haven't been competitive against Intel's offerings, and while the cheaper six and eight-core models do offer good bang per buck, it's only in very specific circumstances where additional CPU cores come in to play. Overclocking them can help, but then you need a half-decent and more expensive motherboard, and the extra power draw compared to an equivalently-priced Intel CPU will likely end up costing you more overall in the long run.
It was also a bit of a gamble taking onboard ATI and integrating its tech into APUs. On the PC side of things, these have actually offered some of the more compelling AMD products, at least at the low end and it has obviously made strides into the console market here too. However, four years after the launch of Llano and the A8-3850 APU
, the range still isn't particularly popular and continues to languish at the low end of the spectrum.
I'm not saying that its APUs would ever have made it to the high-end enthusiast market - I'm thinking out loud here, but there's been little forward progress here, just as there hasn't been with its CPUs. Worryingly, Intel appears to have solved its issue of poor IGP performance too, as early benchmarks of its new Broadwell CPUs equipped with Crystal Well Iris Pro graphics appear to offer a massive improvement in frame rates - so much so that the flagship A10-7870K APU is left trailing.
A cunning move for Intel here would be to release similar graphics horsepower on a Broadwell or Skylake-based Pentium or Core i3 - the likes of the Core i5-5675C are too expensive to do battle with AMD's APUs but drop that kind of performance onto a £100-150 CPU and AMD could see its APU market share hit drastically overnight, especially with Intel's far superior IPC and power efficiency.
While many hoped its latest Fury and R9 300-series GPUs would turn its languishing GPU prowess around following Nvidia's stunning GTX 900-series, they haven't quite captured the hearts and minds of PC enthusiasts just yet either. However, the GPU portion of the company in many ways still seems to be its strongest asset.
Which makes me wonder - is it the only thing that has kept the company afloat and what would happen A) if AMD had never bought ATI and B) what if AMD was to spin off its GPU arm?
One of the last GPUs to be born under a purely ATI banner was the Radeon X1900 XT - a fantastic product so things were certainly plodding along at the company before AMD bought it. However, AMD, on the other hand, was suffering at the hands of Intel with its new Core microarchitecture, following a fantastic period of competition between the two companies. In 2007, Phenom was launched and, well, the rest is history.
The CPU arm of AMD has struggled ever since and it's only hope is that its new Zen microarchitecture, which is due to be released in 2016, can turn things around. Even if it doesn't stand up to Intel's Skylake or even current Haswell CPUs, if there are some serious gains in efficiency and IPC, this could turn things around. Seeing as the likes of the FX-6300 are still very popular, if AMD managed to add 10-20% better IPC to bridge the huge gap with Intel in single and lightly-threaded tasks, it could do wonders for its CPUs, especially at the high end.
Which makes me think that a spin-off - a separation of the companies that merged in 2006 allowing each to focus more intently, could be the way to go. Nvidia's Jen-Hsun Huang put it quite well in a recent conference regarding its Pascal-based GPUs:
'We have found over the years to be able to focus on just one thing, which is visual computing, and be able to leverage that one thing across PC, cloud, and mobile, and be able to address four very large markets with that one thing: gaming, enterprise, cloud, and automotive'
'We can do this one thing and now be able to enjoy all and deliver the capabilities to the market in all three major computing platforms, and gain four vertical markets that are quite frankly very exciting.'
Combining two very complicated technologies from two different markets was always going to be not just tough to do, but tough to sell as well. Even Intel's Broadwell with its much-improved graphics hasn't been received well by enthusiasts so even now, nearly ten years on, the whole CPU+GPU on a chip equation isn't really working, at least not in PCs.
Should AMD go it alone and spin off its GPU arm? Should it stay together and look at ways of improving things? Let us know your thoughts in the comments.