Will the next generation of digital TV make SGI the company that can do it all?/

As the movement toward digital technology becomes more pronounced in all industries, those groups who understand it best — computer hardware and software companies — are actively involved in everything from developing consumer electronics devices to providing “big iron” computing muscle for digital broadcast of movies.

But, generally speaking, the expertise in most personal computer companies of the day, such as Microsoft and Apple Computer, still lies in deep knowledge of the computational requirements for productivity applications such as databases, spreadsheets and word processors. Behemoths like IBM and Digital Equipment are experienced in corporate, or enterprise, computing and understand industrial-strength systems with power and intelligence enough to run manufacturing plants and move vast stores of digital information around a network.

VETERANS IN ‘VISUAL COMPUTING’

As digital media and multimedia develop into technologies to be reckoned with, virtually every hardware and software company is making some effort to learn the intricacies of computing with pictures and sound. Those most dedicated to the task are trying very hard to cover quickly the ground that Silicon Graphics Inc., a computer company based in Mountain View, CA, broke a decade ago. Manipulating and maneuvering 3d graphics, video, audio and other unwieldy data types is imbedded in SGI’s corporate DNA.

SGI was founded (and is still chaired) by former Stanford University computer science professor Jim Clark, who came to Stanford from the University of Utah, a hotbed of computer graphics, in 1974. (Other alums include Pixar cofounder Ed Catmull, Apple Fellow Alan Kay and Jim Blinn, now of Cal Tech and formerly of NASA’s Jet Propulsion Lab.)

Financed by a Defense Advanced Research Projects Agency (DARPA) project to create design tools that would keep the U.S. ahead of the Japanese in semiconductor design, Clark took his expertise in 3d and built what are still the cornerstone of SGI’s products, the Geometry Engine and the Image Engine, for performing the special calculations required to generate 3d graphics in real time.

[Historical note: That very same DARPA project also spawned Sun Microsystems and MIPS Computer Systems. MIPS is in the process of merging with SGI, and both companies are cutthroat competitors with Sun.]

Clark cherry-picked an enormously talented start-up team from watching for the brightest students in various graduate and Ph.D. programs at Stanford University. Then, in 1982, they left to found SGI and together shot the fledgling company to the top of the computer graphics market selling what it calls “visual computing,” and have already leveraged its powerful 3d graphics machines out of the bowels of science and engineering labs into the glamorous world of Hollywood’s special-effects studios.

YEN FOR THE LOW END

For some years now, Clark’s yen for the low end of the market has successfully shoved SGI’s formidable graphics capabilities into smaller and cheaper computers. As the price of high-powered workstations plummets industry-wide, SGI believes its facility with visual computing can finally deliver a useful rendition of digital media into the workplace and beyond.

It’s even official: In April, the company issued a press release announcing that the field of digital media was a “core technology” for the company.

Had SGI made this strategy decision two years ago, it might have a great deal more leverage than it has today. Now even everyone is on the digital media bandwagon; even the traditionally stodgy Digital Equipment is heavily hyping a “multimedia” strategy. Nonetheless, Clark believes that SGI still has the opportunity to be a significant player in everything from delivery of digital television and video programming, via its high-end systems, to providing the quintessential digital media authoring system (via its mid- to low-range workstations), to building the guts of a TV-set-top cable decoder box.

The mirage of ubiquity. Anything is possible, and certainly SGI is in a very good position technically to accomplish these heady goals. But no matter how unique its technical abilities, SGI — like all hardware and software companies, at this point — is in danger of being duped by the mirage of digital ubiquity.

Digital ubiquity presupposes that any company knowing its way around the computer industry automatically has it made in the emerging digital world. But any company that hopes to play in that world — even a company filled with brilliant technologists — has to meditate long and hard upon the fact that it is a dangerous undertaking to move outside of one’s area of expertise.

They’re scratching at the door. In SGI’s case, the risk is two-fold: It has plenty of formidable foes already scratching at the door, hungry both for SGI’s present business and for these new digital media markets it hopes to conquer.

In the future, for example, it may soon become a thorn in SGI’s paw that Bruce Tognazzini, a noted Apple user interface designer, recently defected to workstation competitor Sun Microsystems, followed closely by Bud Tribble, former chief software designer at Next. The folks at Sun are obviously planning a concerted move toward the low end, possibly even the consumer market.

Can SGI do it? Can a company that started as a provider of mega-powerful equipment to the high priests of computer graphics move its wares into the larger world? Given sufficient humility and innate intelligence to search out good partners who know the things SGI doesn’t, it is indeed possible. But as Ringo says, “It don’t come easy” — and that advice goes for all computing companies trying to segue their talents into a world verging on digital.

A FORMIDABLE ARRAY

SGI’s customer base is the envy of the computer business, which is well aware that everyone from Walt Disney Studios to Industrial Light & Magic to the National Aeronautics and Space Administration to Boeing Corp. is willing to pony up significant dollars for SGI’s 3d graphics computers. The company’s formidable product line ranges in price from $250,000 for a SkyWriter “virtual reality generator” to the new Iris Indigo, selling like hotcakes in its entry configuration for $7,995.

The Personal Iris, announced nearly four years ago, was the best-selling SGI workstation in the company’s history until the Iris Indigo. “It was our first mainstream product, and it doubled the size of SGI as a company,” says Jim White of SGI’s Entry Systems Division. Starting at $12,000, the Personal Iris was the first machine to bring SGI’s graphics capabilities to individual users for a price that didn’t cause apoplexy.

Surprised by success. Despite the smashing success of Personal Iris, it is the new Iris Indigo that embodies the changing soul of SGI. Even SGI’s ultra-practical president, Ed McCracken, who by most accounts was at first a little dubious about the concept of a “digital media” machine for SGI’s customers, was surprised by the success of the little purple box, which he claims is already making a profit.

“It’s weird looking, very strange and different,” says McCracken. “But we’ve sold three times more units than we expected to. I was a believer, but still I’m surprised.”

INDIGO BREAKS TRADITION

Indigo broke a lot of tradition for SGI. For one thing, the company is doing a land-office business selling Indigos by direct mail order — something that not even Apple Computer or Sun Microsystems, two of SGI’s competitors for low-end systems, do today.

Second, the company is working hard to snare software developers for the machine. And third, Indigo is the first truly general-purpose system from a company best known for its work in a highly specialized niche of computing, which certainly doesn’t hurt when you’re trying to attract a wider variety of software developers to your platform.

‘Virtual 24.’ Indigo can also boast a rather significant technical breakthrough. Most high-end graphics systems, especially 3d systems, require high-quality 24-bit imaging. Though Indigo appears to provide 24-bit graphics, it does not. SGI’s engineers came up with a system called “Virtual 24″ that simulates 24-bit, 3d graphics using a far less expensive 8-bit graphics board.

This is a not a trivial accomplishment. Typically 8-bit systems can’t do 3d graphics, because they require what’s called double-buffering, which cuts the number of color choices in half. Computer-graphics math says your eight bits would then become four bits, which translates to a mere 16 colors. It is impossible to shade even a single 3d object using only 16 colors.

“Virtual 24 really made a big difference” for Indigo, says White. For example, SGI customer and developer Alias Research, maker of high-end graphics software for the pre-press industry, was really skeptical originally and didn’t believe Indigo would be worth it for the company or its customers. Since then, however, it has released “a bunch” of software on Indigo and is “shocked at the image quality.”

Falling short of perfection. Despite its technical wizardry and come-hither pricing, Indigo isn’t perfect. Though the company has reasons for doing so, it’s slightly ludicrous to read SGI’s promotional materials — “a true digital media-capable system is designed from the ground up for the express purpose of allowing the user easily and simultaneously to work with 3d graphics, video, still images, animation, audio and text … it takes a total integration approach” — then look at the Indigo spec sheet and realize that it isn’t shipped with either a video board or a cd-rom drive.

The IndigoVideo board is a $2,500 option. The cd-rom XA drive is another $1,200. In fact, no storage device is standard. Because of its built-in networking, which means that many of SGI’s existing customers will already have all the necessary software, even a system disk containing the system’s bundled software costs extra.

For now, says SGI, that’s okay. Most of its customers are still from its primary market, engineers and scientists, who want the flexibility of configuring a workstation the way they want it. They say they’ll change the packaging as customers request it.

But as Indigo moves out of familiar markets into terrain covered by the personal computers of the world, Silicon Graphics had better heed the “plug-and-play” call. Such ease of use was Apple’s strong suit right from the start, and if SGI wants some of Apple’s market, it will have to provide at least a modicum of that kind of functionality.

NOW ENTERING THE MAINSTREAM

Despite its noticeable slant toward its existing customer base of software engineers, great efforts are being made to “mainstream” Indigo.

Dave Larson, SGI’s director of software, says a Digital Media Developer Kit will be shipped in July for $495. Carol Peters, former project manager for the Indigo, has shifted jobs so that she is devoted full-time to attracting developers to the platform. So far, more than 20 developers — including WaveFront, Diaquest, Alias Research, Avid Technology and Time Arts — have announced or are shipping software that takes advantage of Indigo’s digital media functionality.

Also on the boards are deals to port AutoCAD to the Indigo, as well as Adobe Illustrator.

A recent developer conference in Boulder, CO, was “incredibly exciting,” says Jeff Barco, who works for Peters. A 14-year veteran of the PC industry, Barco says it’s been very many years since he’s seen developers so fired up to bring a technology to the mainstream.

Indigo QuickTime. A recent deal with Apple Computer will allow suitably equipped, media-capable SGI computers to read and write QuickTime movies. SGI is doing the development on the QT project, which when finished will be shipped free with SGI’s Movie Player and Movie Maker software. To date, Indigo is the only platform that will be licensed to create as well as play QT movies. Many of SGI’s existing customers are thrilled — for them, it will be the first time to allow people without access to high-end SGI machines to see what’s being produced on them.

One way SGI hopes to draw developers is by the sheer influence of the SGI product line. SGI claims the largest binary-compatible line in the industry now, which means Indigo software can run without modification on all of SGI’s computers. Vice versa is a little trickier since audio isn’t standard on SGI’s high-end hardware — a funny table-twist, since features from high-end computers usually migrate to the low end. This time, says Larson, “the features are migrating up.”

Free tools and a new development environment. In July, SGI says it will begin bundling with Indigo (and soon, for all SGI’s platforms) a set of “end user enablers” called the Iris Media Mosaic. These enablers allow users to take advantage of the machine’s digital media capabilities as soon as they buy it.

It’s an impressive bundle, including a sophisticated audio control panel, sound editor and filer, cd and digital audio tape (dat) managers, a video control panel for Indigo Video hardware, a movie maker and player, and Showcase, a powerful presentation package.

“No one ships a presentation package with their systems,” says Larson. “That’s worth a few hundred dollars right there. And there’s hundreds to thousands of dollars more of value that gets thrown in free.”

Also shipped with the system is Workspace, a graphical user interface to SGI’s Irix operating system; Iris Explorer, which uses object technology to snap together a mini-3d application; and a disc called 4D Gifts — “a bunch of unsupported, great little applications that the engineers put together,” says Entry Systems’ White.

A recent deal with Avid may make Indigo an even better deal for digital media developers (for details, see Briefs, p. 21). In short, Avid and SGI are working together on a standard platform called the Open Media Framework (OMF), which allows users in the post-production environment to move digital media data easily back and forth between different tools and computers — a task that is a nightmare in today’s world of wildly incompatible computers and professional editing equipment.

OMF will be ported to Indigo first, via joint effort between the two companies, and it will be SGI’s multimedia development environment for digital media applications, according to Avid’s marketing director, Joe Ricotta. If OMF catches on, SGI hopes that digital media developers will abandon their present development machine of choice — today, the Macintosh — for Indigo, without fear of rendering useless all their previous work.

“People think of SGI as an exotic brand of computer, but it’s less exotic than people realize,” says Larson. “Because of our history, things like the cable business will be an opportunity for us as well, but there is plenty of in-between stuff too.”

JIM CLARK’S ‘TELECOMPUTER’ OF TOMORROW

Indeed, many visionaries in both the consumer electronics and computer worlds have waxed poetic about the coming convergence between televisions and computers. They visualize digital, interactive textbooks delivered to the TV, information retrieval from online databases and libraries, games that link many players in different locations, and the capability to perform telephony functions through this same box — i.e., message retrieval, video telephones and “multimedia mail.”

SGI’s Clark has been beating this drum for a while himself, though with somewhat limited success inside his own company. He’s not alone in his plight. Despite highly publicized moves by such pacesetters as Viacom (see Briefs, p. 23), digital television is not yet here. Those in charge of the bottom line at companies like SGI, with scary competition like Sun Microsystems breathing down its neck, aren’t exactly jumping at the chance to invest millions of dollars in technology that’s not arrived.

A $200 core unit. But Clark absolutely, deeply believes that if SGI acts now, within three years, it can ship a high-powered central processing unit for $200 that acts as the core of a set-top device enabled with SGI’s 3d capabilities. The equivalent of today’s cable box decoder, it would sit at the end of the cable or phone network and deliver output to a cathode ray tube (CRT) screen or digital TV.

“Today’s TVs have tuners in them that are totally unnecessary,” says Clark. “There’s potential for a new industry here that would build such a CRT.” In an attempt to rally his industry around such an effort, Clark wrote a paper called “A TeleComputer” for next month’s Siggraph conference (see Events, p. 28), in which he outlined the essential qualities of such a module.

FOUR BASIC TECHNOLOGIES

What such a device would require, he says, is four basic technologies: digital video, digital audio, and the decoupling of both transmission/reception and resolution. This means a digital TV’s set-top decoder box must be smart enough to figure out how to display a digital TV frame or datastream, no matter what kind of screen was attached to it. (This quality is also known as device and/or resolution independence.) Such decoupling is good for many reasons, including easy adoption and display of high-definition TV signals when they arrive.

At the core of the “teleprocessor,” as Clark calls it, would be a very high-speed (500 million instructions per second) RISC, or reduced instruction set, central processing unit, a real-time operating system (which he doesn’t feel yet exists), high-density memory, very high bandwidth (6.4 gigabits) networking for input and output, and advanced 3d graphics capabilities. Encryption and decryption capabilities, he says, are also essential components to allow consumers to pay for services.

To keep costs as low as possible, Clark believes it may be necessary to circumvent intermediate parts suppliers. In fact, he says, “success might even require that a single company be able to integrate almost everything, from semiconductor fabrication to the telecomputer itself.” This is no surprise, considering the company’s recent merger with its microprocessor supplier, MIPS Computer Systems (more on MIPS, below).

‘THERE’S MONEY TO BE MADE’

Though its main role will be delivery of entertainment — most likely candidates are digital movies and
television programs — Clark believes that such a device would start a rush of interest in the next wave of interactive games. “People always ask whether the next generation TV really needs 3D,” he says. “Some say no. But digital video alone may not be enough. You may need games. You may want to get virtual reality into the home without the head-mounted displays.”

Clark contends there is a “huge market” for a realistic, multiparticipatory, interactive flight simulator. “Flight simulators are a compelling thing. You can call a buddy and play the game, have a little contest, gang up on other people. I’d like to supply the game software,” he says with a smile. “There’s money to be made there.”

Sluggish activity level. Clark is very hot on building this device, but he admits that at SGI, the scale of activity for the project on a scale of one to 10 is “about three.” President McCracken says that’s partially because SGI never plans things more than 18 months in advance. “It’s one way we stay close to what’s happening in the industry,” he says. So Clark is out on the rialto, pushing hard to get partners lined up behind the concept. One surprising snag he’s hit in the consumer electronics world is some heel-dragging on the part of some large companies that, as he says, “don’t particularly want to see analog video go away.”

A $10 BILLION SERVER BUSINESS

Though Clark’s focus today is mostly on the consumers of the near future, he’s also interested in where SGI’s high-end computers can fit into the loop for delivering information to those set-top decoder boxes.

He believes that SGI’s slant toward visual computing makes it a natural solution for delivery of picture information, no matter whether the delivery vehicle is phone lines, cable or direct-broadcast satellite. Though he declined to give specifics publicly about what companies he’s talking to about such potential solutions, let’s just say he is pushing ahead.

‘Client-server’ works for TV, too. Clark says it’s an easy map from the “client-server” model of today’s computer networks to what the cable and phone companies will need from a hardware provider. “The client in the consumer environment is simply the digital television in the home,” he says. “The server is a bunch of stuff at the headend or the local exchange. Properly mutated, this is simply a multimedia computing system.”

He believes such servers represent “a $10 billion business opportunity” for SGI. “Look at the local loops,” he says of the phone companies. “They’ll each need $4 million worth of gear to do this. And how many local loops are there?”

Though Clark doesn’t see SGI doing the installations itself, he sees a land office business for a system integrator in putting such digital television systems together for media delivery companies. And speaking of integration, Clark also believes that such a scenario, carried to its logical extreme, might show that compatibility between the client — i.e., the digital television — and the server at the headend/local loop might just be important.

WHERE MIPS FITS IN

Voilˆ! Enter the MIPS Computer Systems merger with SGI. MIPS, long plagued by bad management and good technology, is the designer and seller of the RISC processors upon which SGI’s entire product line is based. SGI decided in early spring to “merge” with MIPS — resulting in a $200 million stock swap.

As a result, SGI has more than a few months of rough going as it figures out how to merge MIPS Computer Systems into its operation. So, despite its success story with Indigo, the company is cash-poor at a time when it most needs to whip up support around its newfound religion of digital media.

Tactical value. Both companies’ stock dropped as a result of the announcement, so the jury is still out on whether the merger will be a success. But both Clark and McCracken agree that in the long run MIPS will be a great asset to SGI.

“There is great strategic and tactical value of owning state-of-the-art technology,” says the pragmatic
McCracken. “Half the value for us is simply in the microprocessor itself and access to the technology.” And Clark obviously sees MIPS chips at the heart of every one of his future set-top digital decoders, a massive potential market if he can pull it off.

WHAT A SANDBOX!

As in the case of Bill Gates and Microsoft, never underestimate the potential of a company that has bright, curious and ambitious engineers at the helm. Though Clark is often painted as a kind of wild dreamer, his ideas for SGI’s future are certainly not far-fetched in a world where cable TV researchers are actively and publicly searching for ways to deliver digital television.

All but one of Clark’s hand-picked professors and graduate students from the early days are still at SGI. They’ve obviously found a sandbox that’s more fun to play in than anywhere else. If Clark and McCracken can provide them with the proper direction in terms of technical problems that need to be solved, and sufficient resources to do the job, their track record to date makes a pretty good case for success.

After all, not many companies could realistically consider supplying equipment and expertise for everyone from the cable headend to the film, video and TV post-production houses to the interactive multimedia producer and finally, to the consumer’s own TV set.

BUT FIRST, A REALITY CHECK

But they do have some prejudices to overcome, and some realities to face.

First, and absolutely mission-critical, SGI has to figure out a way to move past its deep identification with the science and technical communities. To abandon its core customers in the most lucrative market going — companies that need complex problems solved, and damn the price tag — would indeed be foolish. But the economics and customer needs of the cable business on one end, and the consumer business on the other are radically different, and SGI cannot expect them to adapt to the way SGI does business.

Second, the company must pay more than lip-service to making digital media easy to use and incorporate into products. It is vital that the company adopt a single, standard user interface for application developers. “Developers are begging for it, but it’s hard to get the company to do it,” says one SGI employee. This is foolish and counterproductive.

As producers are finding out with no small degree of pain, it is far from easy to develop a multimedia title. It is also insanely expensive. Potentially, SGI could do much to win developers if it wasted no time in standardizing a user interface (ˆ la Macintosh) and development environment, and put together a standard plug-and-play Indigo for those solo artist-types who will really be the ones to put digital media on the map as a new entertainment and/or communications medium.

Leverage what you’ve got. Next, SGI cannot bank on its stellar reputation in the film production industry to buy it anything special in the distribution world. But it should find a way to leverage its existing, enormous cachet in the entertainment business. For example, if it really wants to be a supplier of head-end equipment for digital television, SGI ought to get its foot in the door before any big deals go down. There are more than a few opportunities to get in there and muck around with the problems that need to be solved.

Clark’s instincts are that by the end of 1992, everyone will be whipped to such a pitch about digital television that there will be lots of investment money available. But there’s an awful lot going on right now, today, in the subterranean grottos of cross-industry deal making. SGI can’t afford to wait any longer if Clark really wants to move down from the rarified air of SGI’s elite market and muck around in the world at large.

Denise Caruso