Gah, the "Nor is a tablet..." through to the "...failed merger" part was meant to be italicized; I must have goofed on a formatting tag. Cheers.

But the success of tablets and smartphones and the blurring of the line between netbooks and small laptops has shown that one doesn't need an Extreme Edition CPU to do homework, Facebook, YouTube, and light gaming.

Right, and tablets and phones will be fine for a majority of users going forward. I'm 100% OK with this - conventional workstations are likely to become more expensive, but there will also be fewer "dumb user" concessions made in their design. That doesn't mean tablets and smartphones will be well-suited to ArcGIS, PETRA, or video transcoding. I'm also not fond of the carrier/manufacturer-mandated, inherently boxed-in nature of the experience.

Power constrained? Make it quad core, with two cores turned off while not connected to a charger. The backlight uses a huge chunk of the power anyway, and a docked PDA or phone can run with the screen turned off. This behavior could even be advertised as the return of the turbo button [wikipedia.org]. Space constrained? So are ultrabooks, and the solution is the same: dock to an external hard drive.

Yes, but as admirable as per-watt performance is, ARM's absolute performance offered still hovers somewhere around the Pentium II level in best-case scenarios. Toss in some dedicated-purpose DSPs and that can be mitigated, at the cost of some general flexibility. The focus on miniaturization does result in a hobbling of speed, expandability, and general flexibility, and as some savvy others have pointed out, will probably result in the eventual, painful reinvention or reinterpretation of certain desktop metaphors when a portable device is effectively "docked."

Nor is a tablet. Nor is a PDA such as Apple's iPod touch or Samsung's Galaxy Player. Tablets and PDAs use Wi-Fi, which most often connects to a wired last mile, and no wired ISP that I've heard of has run the sort of subsidy model common in the North American cellular market since the dial-up days of i-Opener, WebTV, and PeoplePC. Nor is an unlocked GSM phone, now that (as I've read) AT&T has given up some spectrum in a more commonly used band to T-Mobile USA as compensation for the failed merger.

But it IS limited by what the manufacturer or designer intends for it to do. There will be a market going forward for an expandable, upgradeable PC, even if it shifts back to a more professional focus. My wife's iPad is a joy to use for lots of casual computing purposes, but for my kit I know where I plan to spend my money.

A separate desktop PC can be leagues more powerful than a cellular phone because it's not hobbled by the same frugal power or space constraints. But at least as importantly, a more conventional PC isn't tied to a carrier, with all the baggage - privacy, monthly billing for service, &c. - that entails.

Just curious - did you happen to try disabling network connections, disabling firewalls, using the full-size network installer from a local volume, rebooting, then re-enabling firewalls before enabling network functionality again?

Don't forget: as of the current preview build of Windows 8, you also can't add tablets to a domain. So they're REALLY shooting themselves in the foot by assuming that tablets are only consumer-level devices that will never find a footing in business.

A refutation followed by an assumption of fault is not a valuable contribution to a discussion. I shouldn't burn karma responding to ACs, but what were you hoping to accomplish by thumping your sad little e-peen here?

XP will be around for a long time in various capacities - probably at least as long as MS-DOS has lingered in point-of-sale terminals and niche industrial roles. Directed I/O in virtualization is a godsend for users of a lot of older lab equipment - now it's possible to run a 64-bit host OS with an XP VM that has unfettered access to, say, an old PCI controller card for a transmission electron microscope which hasn't had driver or software updates since 2003. I've never understood that missionary zeal that people develop vis-a-vis old software - use what works, in whatever combination works best.

Assuming that you're using vanilla Doom 3 content, that may be true. But if the Doom 3 mod community receives a shot in the arm as a result of the source release (and I don't imagine there's any reason it wouldn't), it's likely that maps and content will be released which push modern hardware quite a bit further than the original game ever managed. At that point, every speed optimization available will make a positive difference, especially if map and/or asset complexity increases by some significant factor.

Note that this is probably more academic than many here realize - as long as the offending algorithm is not "officially" released, I'm pretty sure Creative won't care enough to go after individual projects with a legal banhammer to force them to stop using "their" algorithm in non-profit projects. And if worse comes to worse, I'm 100% sure someone will implement shadowmap support within the first year or so, which will render* this issue irrelevant.

* Oh, that's a pun. I'm sorry.

...but on the bright side, Carmack's Reverse can always be reimplemented by the open source community, and this may hasten the implementation of shadowmaps in source ports to replace Doom 3's stencil shadows. The visual quality improvement would be non-trivial, and for all but hardware that was *ahem* not exactly groundbreaking when Doom 3 was new, the performance delta shouldn't be too massive. In some scenarios it may even be faster.

The aptly named SleepResearch_Facility's album Deep Frieze is a great place to start.