MacEdition’s Soup: Click here to return to MacEdition Home Page

So Sayeth Soup: Tossing X Into The Seething Sharkpit of x86

By SoupIsGood Food, (soup@macedition.com), 16 November 2000

In the same insanely fertile world that can incubate a battery-powered and hand-held Atari 2600 clone, complete with faux woodgrain, the suggestion that Apple may cheerfully glom onto x86 seems downright plausible. There’s a lot of excitement over the possibility, too.

Okay, not so much excitement as orgiastic hysteria, mostly coming from hardcore geeks with a deep-seated PeeCee envy. Mostly the lust is for cheap, generic equipment with a buzzword-compliant processor and graphics card, and these fervent desires change according to who’s infinitesimally ahead in this week’s benchmark pissing match. Athlon, Pentium 4, Nvidia, Voodoo, whatever. Wait a month and you’re already hopelessly obsolete. You can’t follow the game without a scorecard, and the scorecards often argue endlessly with each other, using whichever benchmark that currently tells them what they want to hear.

The assumption with this cheap and/or powerful PeeCee gear is that all the parts fit together like Legos, and you just flip a switch and boot up into a happy Mac face. It don’t work that way. Your crapmatic generico motherboard may not like your el cheapo network card which may toss your name-brand, big bucks graphics card into a tizzy.

Building PeeCees is a time-consuming and often frustrating hobby, just like hotrodding cars. “Bolt On Performance” never is. After you get it bolted on, you gotta massage and tweak and finagle things until they work right. This is why companies like Compaq, HP and Gateway exist: building a PeeCee from parts is no fun unless you want a new way to spend a rainy weekend afternoon.

Apple makes most of its moolah from hardware sales. The average margin on the iMac alone is four times greater than the selling price of Mac OS X Beta. If Apple couldn’t compete with the PowerPC Mac cloners going for its jugular, it will be minced to shreds in the hypercompetitive PeeCee market. There are a hundred times the players, not to mention bargain-basement generic specials at the local Mom-and-Pop screwdriver shop. Once the Mac OS goes x86, Apple can kiss its hardware profitability goodbye.

There’s really no compelling advantage that Compaq or HP or Gateway have over Apple’s offerings. Apple provides a better user experience than any of the PeeCees out there, and the current crop of Macs are very cost-competitive with the brand-name systems. Macs these days may not be as sturdily stolid as they were in the days of yore (I still have a Mac Plus going strong, fourteen years after the fact) but they are every inch as good as the top-of-the-line systems rolling out of the big-name PeeCee makers, if not better.

Well, there’s one advantage the PeeCee crowd can claim – speed. Megahertz for megahertz, the Mac outstrips the PeeCee. Unfortunately, the PeeCee’s got a whole heckuva lot of ’em hertz thingies with more on the way. Motorola’s R&D pace of late can best be described as glacial, and IBM seems uninterested in picking up the fumbled G4 ball and running with it. Which is a shame, as it hurts more than Apple – the G4 is one popular chip in the high-end embedded market. Companies like Mercury and Sky were forced off of Intel’s extremely innovative i860 RISC processors in the late nineties due to vendor indifference, and they will probably not be pleased at the prospect of having to do it all over again. Here’s a hint: they probably won’t be going to Motorola’s ColdFire or IBM’s G3 variants.

So, to keep their embedded market happy, one or the other is going to have to budge. Either IBM or Motorola or both will probably get on the ball and crank out competitive G4s one of these days. Even so, the G3 and G4 aren’t that much slower than the x86 competition right now. It will take a lot more firepower than either AMD or Intel has on the horizon to radically outstrip the Mac, even at a lousy 500mhz. Emulate PowerPC Mac Classic apps at native speed on an Athlon at 1.2 gigahertz? Only in the darkest, IRQ-conflicted dreams of the most deluded Mac haters.

More to the point, when you look three years down the line, PowerPC has got legs to it, despite its current falterings. The next-generation PowerPC processor is already here. IBM’s Power 4 is nothing short of a performance monster, lapping both the legendary Alpha processor and Sun’s new UltraSPARC III chip. At eight grand a pop, don’t expect to see it in a Cube anytime soon – but it’s a clear indication of what the G4, and whatever comes after it, can achieve.

Meanwhile, Intel’s next-generation chip family, IA-64, is more than two years late and will be dog slow and unusably buggy in its first version, the Itanium; even the Pentium 4 is likely to be faster. The next version of the IA-64 chip, designed by Hewlett-Packard and called McKinley, won’t have quite so many of the same problems plaguing its little brother, but in the meantime good old RISC keeps surging ahead, breaking one speed record after another. AMD’s Hammer, a 64-bit x86 chip in the works, is actually pretty good, if way overdue. The problem is that it currently has next to no industry support, and will die a long and lingering death without some serious developer commitment.

In the long haul, we are still better off with a PowerPC strategy than an x86 one. IBM and Motorola both have a lot riding on the high and low ends of PowerPC RISC architecture. Apple would be better off using its time, money, PowerPC patent portfolio and painful noogies administrated by Steve Jobs to convince its wayward AIM partners to bend a little consideration to the Mac in the middle.

The Unix hardware market also aptly illustrates why we shouldn’t rush into PeeCee hell. Larger and smarter companies than Apple have dashed themselves on the “industry standard” rocks, leaving them high and dry while competitors with “closed” architectures have prospered.

Take Silicon Graphics, Inc. for instance. Now just called SGI, it is a large Unix hardware company known for its servers and sexy graphical workstations. SGI did its best to move to a non-proprietary platform by lining up behind Intel. It shipped Pentium workstation systems and informed all its developers and customers to expect a switch to IA-64. They even went so far as to spin off their RISC microprocessor design division, MIPS (now a very successful independent RISC processor company), in preparation for a future that never came. IA-64 isn’t here, and SGI’s Pentium-based workstations are a failure. SGI’s developers and users lost faith in the company, and have been abandoning it in droves for the past four years, leaving this once-mighty mover and shaker a washed-up has-been.

In the meantime, Sun, another major Unix hardware manufacturer, stuck to its proprietary guns. It built bigger and better systems, and never faltered with its “closed” design and product strategy, despite the slow pace of development on its new processor, the UltraSPARC III. Not only is Sun surviving in the age of Wintel, it is absolutely thriving. Its low-cost workstations, multi-million dollar mainframes and everything in between are all based on its own RISC technology and decidedly non-industry-standard systems.

Even if the PowerPC winds up an evolutionary dead end for the desktop, a move to x86 instead of another proprietary platform would still rank up there with monumental business blunders like using doped cellulose material impregnated with aluminum on lighter-than-air craft coverings, or launching a line of luxury cars in the middle of a recession.

The PeeCee marketplace is a graveyard of operating system also-rans. Microsoft chews them up and spits them out because x86 is Bill’s yard, never forget it. He’s danced on the bones of DR-DOS, GEM Desktop, Desqview/X, Geoworks, OS/2, BeOS, SCO Unix, and, yes, OpenStep. What do all these operating systems have in common? They were wildly innovative, light years ahead of Microsoft’s offerings, and in Bill’s way!

Take for instance the cautionary tale of OS/2. IBM’s OS/2 really was revolutionary – a modern OS that was blazingly fast, amazingly stable, and with a clean user interface that beat, hands down, anything Microsoft could come up with. It even ran Windows 3.1 and DOS programs better than Windows could! IBM poured some serious PR and marketing muscle into this winning product, with a series of widely aired television ads and a full-court press in the trade publications. Big Blue’s well-connected sales force made some serious inroads into banks and other financial institutions, and set the stage for major growth. The largest, most successful computer company on the planet, with a product that was affordable, powerful and well-supported, entered the PeeCee arena with every intention of carving out its fair share of the market.

Microsoft ate them for lunch.

OS/2 is now just a fast-fading memory, because Microsoft used every low blow, every dirty trick, and every illegal little bit of leverage it had to run it out of the x86 market. The first time Microsoft went head-to-head with antitrust law was over the dead body of IBM’s wunderkind. Every commercial OS competitor on the x86 platform, in one way or another, by hook or by crook, has been shut down or marginalized into nonexistence by Microsoft.

This is the seething shark-pit the drooling fanboys want to drop Apple into? Where IBM, Big Blue, the company who invented the PeeCee, couldn’t sell a superior, well-marketed OS to save its life? Sounds more like a plan for assisted suicide rather than a road to renaissance.

These days, only Linux and FreeBSD Unix operating systems enjoy some degree of continuing success on x86. This is because:

  1. the hardest of the hardcore PeeCee nerds miss MS-DOS and the command line,
  2. all CompSci programming classes are taught using Unix of some stripe,
  3. it’s free, and
  4. it’s propped up by religious zealots who really don’t care how much the product sucks, so long as they know they could, if they wanted to, read and rework the source code. (Although most of them can’t, won’t, or just don’t.)

The kinds of people who run Linux simply couldn’t care that the Mac offers power to the users through the interface, or that the UI is orders of magnitude above and beyond the quality of their miserable little X Window toys. If they did, Be would be leading the alternative OS pack instead of sinking into irrelevancy. (Be Inc. no longer has any hope of making a profit on “industry standard” PCs, and is desperately looking for a way out and onto a special “internet appliance” platform.)

Typical free-Unix enthusiasts only care about fronting for their geek buddies, to prove they can run with the nerd herd. They enjoy the arcane complexity, the neverending parade of esoteric shared library conflicts and bizarrely-named utilities and applications. They confuse “hard to use” with “powerful”. Matter of fact, most Linux users have their computers set up as a dual-boot system, with the option of launching either Windows or Linux at startup. They drop back into Windows from their beloved free Unix variant whenever they want to use a computer instead of tinker with a hobby.

Is this the ultimate fate of the Macintosh? Just another geek toy, to be tossed aside for Windows whenever there’s real work to be done? It will be if Apple ports Mac OS X to x86. The soul of the Macintosh is what you are trying to sell when you sign this dotted line.

E-mail this story to a friend