Walker's recent post on the prima facie petrifying plummet in PC sales got me thinking. Or rather rebooted a thought process I've been mulling lately. Just what is happening to the PC? You can make a strong argument, for instance, that we're entering a golden age of PC gaming. Faster graphics, cheaper ultra-HD screens, the new consoles as thinly disguised PCs, VR technology - the next five years or so are going to be fabulous. But there are also signs the wheels are falling off the entire enterprise of the PC as a computing platform. Then there's the ever-present threat of Moore's Law hitting the wall. Can we say anything concrete about it all (the future of the PC, not the looming wall)? Prepare yourself for a multi-topic treatise...
Just to address that PC sales thing, my take is that the current noise involves sales of PCs we don't care about. That's where the suffering is. Feeble client boxes for corporate drones, cheapo home PCs sold predominantly on price, that kind of thing. Instead of buying a crap desktop, people started buying crap laptops. Now they're buying crap tablets.
Actually, did you know 225 million tablets will be shipped out in 2013? Combined desktop and laptop PC sales are put at around 300 million. For 2017, those numbers are expected to be 400 million and 300 million respectively, with desktop PCs being the only shrinking market (by about 10 per cent over that period, if you're wondering - as ever, all IDC numbers).
Think happy thoughts
The main thing that hits me is just how huge those figures are. I get a bit giddy thinking about the rampant consumerism, the scale of the raw materials usage, the environmental and human costs. Is it really sustainable? Must...keep...thinking...happy...thoughts...!
Anyway , I shall suppress the darkness and add a little context. The best figures I can find put the combined total of Xbox 360 and PS3 sales at around 150 million. That's total sales to date since those consoles were launched, not annual sales.
So, all the evidence is that performance-orientated PCs for gaming are in rude health. As far as I know, GPU sales – a key metric for PC gaming – are stronger than ever. And as John said, Steam continues to be the most virulent strain in the global Petri dish of gaming – user numbers up 30 per cent up in the last year.
Then there's the other highly relevant angle regards consoles. The arrival of new models from Sony and Microsoft. Architecturally, they're almost pure PC, pulling the three major gaming platforms closer together than ever.
For the consoles themselves, that's not immediately of much import. It's mainly technological expediency. It's the cheapest, easiest way to hit performance targets. For the PC, however, it could prove significant. At the very least, you'd think that porting games across will be much more straight forward.
High performance ports
The performance advantage of a high end PC means running console ports was never going to be an issue for serious desktop machines, of course. But the close relationship between the latest consoles and the PC will probably help things like cheap PC laptops and tablets with integrated graphics become gaming viable sooner rather than later.
Likewise, those new consoles will inevitably broaden the general game development horizon. Forget all the shader and core counts. Having 8GB rather than 512MB to work with will make a huge difference.
As for PC hardware itself, this is where the message gets a bit mixed. The graphics technology development cycle has gone a bit weird, partly thanks to Nvidia initially choosing to not put its most recent high GPU (GK110 aka Titan) into PCs. And the rate of development seems to be slowing. But really roughly, we're still looking at refresh cycles every two years or so.
The CPU side is a lot less edifying. AMD really has failed to deliver over the last five years and Intel has taken the opportunity to sit back and coin it in the meantime. Mainstream CPU performance (I barely count Intel's LG1366 and LGA2011 platforms as PC tech, it's really server and workstation) has only crept up in that time frame. And unless AMD's HSA thingie does something spectacular (unlikely!), I don't see much reason for that to change in the next five years.
Still, you can make an argument for the CPU side of things being good enough. So, let's recap. We've got gaming PC sales and PC gamers themselves rocking on and in numbers, the important graphics part of the hardware equation pushing on, the latest console tech tending to magnify the performance of PCs and good reason to think a new wave of technical innovation on the game dev side is on the way.
Oh, and lest ye forget, with Oculus Rift and others edging closer to retail reality the next five years should also see VR technology finally become viable. Big, high def screens are only going to become more affordable, too. Ditto large solid state drives. Mix that all up and I reckon you have a recipe for a golden age of PC gaming over the next five years. And then what? And then Moore's Law, that's what.
Give it up for Gordon (Moore)
Just a quick note on what Moore's Law is for the uninitiated. I'll quote myself from a piece published elsewhere because I'm pathologically idle and it seems to do the job.
“It's simply the observation that transistor densities in integrated circuits double every two years. Put another way, Moore's Law says computer chips either double in complexity or halve in cost – or some mix of the two depending on what you're after - every couple of years.”
Now, Moore's Law is a bit like peak oil. Every time somebody predicts the oil will run out or Moore's Law only has a few more years left to run, somebody finds more oil and the wizards in the chip fabs find a way to keep the party going.
And yet the fun must end one day. Oil is probably a bad example because it can be synthesised from sunlight, water and atmospheric CO2. But I digress, the point with Moore's Law is that we are fast approaching the limits of physics as they apply to current chips.
Prophet of doom
You can read an interview with the latest doom monger, a chap from Broadcom here. But the basics go something like this. Today we're at about 20nm regards production technology and transistor gate size.
When we hit 5nm, the gates will be just 10 atoms wide and that may well be the limit of functionality. Even if you could shrink a transistor down to an atom or two, that would still only represent a stay of execution. Based on transistors, there is a physical limit to computer power as we know it.
The guy from Broadcom also thinks things have already fallen off when it comes to commercial viability. It's becoming crazy expensive to tool up the latest production tech. For me, the analogues here are jet travel and human space exploration.
The fastest passenger jets are now much slower than the best from the 1970s. The absolute horizons of human space travel have shrunk, too. That's not because we can't fly faster or put a man on Mars. It's because we don't fancy it. It's too bloody expensive and nobody cares enough.
The same may apply to computer chips in coming years even if the physical barriers don't come into play. Of course, a huge paradigm shift – perhaps quantum computing – may pop up. And there's plenty of scope for clever coding to extract the equivalent of at least a couple of Moore's Law cycles out of chip performance through more efficient software. When the hardware stops getting faster, that will really focus minds!
But my feeling is that computers – at least the ones we buy and use ourselves – won't keep getting faster at anything like the rate we're used to. Again, the transport analogy applies. For a while, further and faster seemed inevitable. But it didn't work out that way. The assumption of ever faster and cheaper computing may suffer a similar fate, at least for a time.
For the PC specifically, there's also the Windows angle and Microsoft's ever-present ability to alienate the installed base. Windows 8 is not encouraging. Given how fast Android devices are multiplying and the prospect of mobile ARM chips closing the gap with x86 processors for performance over the next five years, you'd think the Wintel alliance of x86 processors and the Windows OS needs to be on top of its game to stand a chance.
All of which makes for a bit of a ramble. But really my message is pretty upbeat. Enjoy the next five years. It's going to be spectacular for PC gamers. Just don't assume it's going to last forever.