The USS Quad Damage

The PC is dead. Long live the PC

We can look at death as the end, or we can look at it as a transition.

We have a whole new kind of PC waiting for us, one we've wanted for 20 years

People are talking more and more often about the “death of the PC”, whatever that means. Both “death” and “PC” are things so ill defined in this context that the prediction makes no real sense. However, I’m going to try and divine meaning from it. Because you see, I think the discussion says something about the changing landscape of computing — and I think the computing landscape is changing.

When most people talk about the “death” of the “PC”, what they mean is that no one in the future is going to buy Windows Desktop computers, ala Dells. Some mean to go further still, implying that even laptop computers will be replaced by the new gods of computing — tablets, smartphones, and perhaps netbooks. Everything will be “done” in the “cloud” and everyone will laugh at people who use keyboards — sorry, “obsolete UI paradigms” to do “work”.

Of course, the stalwart and conservatives believe that “Windows Desktop Computers” aren’t going anywhere because that’s where any actual work still happens, and only hipsters and managers use tablets: where the hunt and peck typist is king; and all they are good for is browsing the internet all day: a place where pretentious rich kids retweet #kony2012 to try in vain to save “Starving Africans”.

Neither group is too far wrong. The “new computing paradigm” is about devices which waste time, where the “old computing paradigm” is about devices to do work on. However, people have more money to waste time on, and less to do work with. Dell can’t sell any more desktops or laptops because the ones that exist are good enough, and people buying iPads are enough in number that the sheer economies of scale can embue a brain-dead product with enough “actual work” applications.

However, there is another element to the argument, and it has to do with people — PC people: the tinkerers, the doer-upperers, the min-maxers, are left out in the cold. If not enough people buy the old desktops, the desktop parts become more expensive. If they become more expensive, then why spend all the money to have the best system out there when a cheaper iPad will still have better performance? I think the sneering and anger from “PC People” is part of the whole “shock, denial, anger, acceptance” cycle.

But fellow PC people: I have good news. Once we get to “acceptance”, we have a whole new kind of PC waiting for us, one we’ve wanted for 20 years. You see, the x86 architecture, as much as we love it today, has a complex history which has left it with many scars. Whether it’s the legacy of “640k ought to be enough for anyone”, or the transitions from ISA to PCI to PCIe, leaving memory mapping holes peppered throughout the system, or not being able to use features which newer bus types afford, or even the “I have 4GB of RAM but 1 GB is wasted due to inefficient memory mapping”, this is a system which we’ve loved as we’ve hated it. And that’s without even asking for the Unified Memory Architecture or other fancy features, something reserved for bourgeois SGI systems and other supercomputers.

But shed the x86 and we’re in a world of possibility. The Raspberry Pi, a $40 odd computer, something which could realistically be used as a “second PC” for many “PC People”, could show us the way the PC could truly be. Sure, the 256 MB RAM is “cute”, and the processing is anemic by PC standards, and there’s no expandability to speak of, but treat it like the 8086 of our time, and things start to look clearer.

It has a true Unified Memory Architecture, because the graphics are on-chip. Speaking of graphics, for an “8086”, this thing is a beast. If you could get OpenCompute running on it, it might equal the fastest PC out there (sans graphics card) for certain tasks. It’s a RISC processor and sips power, has the ARM instruction set, and still runs everything your linux box can run. What’s not to like?

Well, expandability for one thing. This thing can’t take any additional RAM and has no PCIe slots to speak of, but think about it: is that really a bad thing? I’ve found that every upgrade I do, I need a new CPU, mobo, and RAM. I’ve probably upgraded my RAM about twice in my life without upgrading anything else, even though I plan out extra RAM slots for every PC I build. As for PCIe, take a look at your PC. I’d think the only really performance sensitive thing on the PCIe bus is your video card. In an age when even PCs are moving towards APUs, having a high speed PCIe bus could easily be put in the “optional” category. The only thing the Rasp. Pi could do with is SATA.

The story on the CPU side also looks quite good. The Snapdragon S4 will probably have similar computing power to an i5. This doesn’t sound too great until you realise that they were trying to keep power draws to a minimum. Take the S4 and re-design it for speed, and we may be looking at power similar to best-in-class PCs. Keeping in mind the fact that even NVidia are in the ARM game now, as are AMD/ATI, and it certainly makes the situation... eyebrow raising. MIPS, too, could hit this space. MIPS is already 64 bit, and offers some very high performance stuff.

All it’s going to take is for enough interest to gather that someone makes a “motherboard” for ARM PCs. I’m guessing / hoping it could be something a little like this: ARM or MIPS CPU 4 – 8 cores with maybe 256 / 512MB RAM on chip. This should ideally be 1 or 2 Wait States. External RAM support on-chip, maybe the standard 7-8 Waits and single channel external memory. Then a south bridge with some PCIe 4x and SATA. Something like this could start a new PC revolution. Hell, it would even run Windows 8.

I know this is pie in the sky; wishful thinking. Maybe none of this will happen and either the PC will die a slow and protracted death, or maybe it will make a resurgence in its current form. But for me that would be a shame, because we could see the dawn of a new age for PCs. And it would be worth the wait.