Computer Chronicles Revisited 94 — COMDEX/Fall '87

The second annual Computer Chronicles episode to focus on the fall COMDEX show in Las Vegas aired in November 1987, just a few days after the event concluded. Back in the San Mateo, California, studio, Stewart Cheifet and George Morrow looked at some video footage recorded on the show floor. Cheifet noted this was the biggest COMDEX ever, with the most exhibitors and attendees. But what went on at the event? Was there any buying or selling actually taking place? Or was it just PR and gawking?

Morrow said it was neither. COMDEX was certainly a social event. But it was really a networking event more than anything else, where you got a chance to see everything that has happened and will be happening for at least a little white. You got to renew your acquaintance with everybody in the industry–to look that person in the eye you wanted to buy Winchester disk drives from and ask how good they really are.

IBM’s OS/2 Announcement the Talk of COMDEX

Cheifet narrated the first remote segment from the floor of COMDEX. He reiterated this was the biggest COMDEX show ever, with over 1,500 exhibitors and 100,000 visitors. One notable returning exhibitor was Apple, which had shunned COMDEX in the past. While the company had no earth-shattering products to announce, its showroom was constantly packed with curious crowds.

For its part, Cheifet said that IBM occupied its traditionally vast floor space at the Las Vegas Convention Center and Big Blue used the event to announce a December 1, 1987, shipping date for the OS/2 operating system. Indeed, IBM’s PS/2 computer line was indirectly the focus of this year’s COMDEX, as industry observers speculated on which company would be the first to introduce PS/2-compatible chipsets and motherboards. While several manufacturers offered compatible plug-in cards for the PS/2 line, few were willing to commit themselves to more than vague predictions.

Morris Jones, Jr., of Chips & Technology told Cheifet there were several reasons for this. First, the PS/2 was a very complicated product to clone. It was taking companies longer than expected. The BIOS in the PS/2 was very large and there were approximately 70,000 gates of IBM design logic on the motherboard. Second, the PS/2 marketplace hadn’t grown real significantly. It would likely take three years before the PS/2 line dominated the market in terms of volume.

Cheifet said that just below the surface of any discussion about the PS/2’s microchannel architecture (MCA) and any potential clones was the legal issues–nobody wanted to be sued by IBM. Jones said his firm had discussions with IBM and he believed the company would maintain a fair policy of licensing all of their utility patents related to the PS/2. But IBM would also be “very onerous” about any copyrights in the machine’s BIOS. There was always concern when dealing with IBM, Jones emphasized, as they had more corporate attorneys than most companies had employees.

Would-Be PS/2 Clonemakers Worried About IBM’s Response

One potential PS/2 clonemaker, AST Research, was attempting to satisfy both IBM and customers, Cheifet noted. AST demonstrated its 386 machine with “smart slot” architecture. AST claimed that by side-stepping the CPU during data transfers, smart slot provided the multi-master functions of IBM’s MCA while maintaining compatibility with older PC-AT machines.

Another potential clonemaker, Western Digital, displayed compatible motherboards for the PS/2 models 30, 50, and 60 at COMDEX. Cheifet noted that Western had not announced a formal release date for these boards yet claimed they would be 100-percent compatible and original. Roger W. Johnson, the president and CEO of Western Digital, told Cheifet that their boards used Western’s own devices for core logic, VGA displays, hard disk and floppy controllers. Johnson claimed that “serious” OEM manufacturers were interested in the Western boards to build PS/2 clones.

But Cheifet cautioned that Western Digital’s leap into the uncharted waters of PS/2 compatibility was bound to attract the keen scrutiny of IBM’s legal department. Johnson retorted that Western had never used–or ever intended to use–IBM’s proprietary information. He said that IBM claimed certain patents on the PS/2 but would not identify them specifically. Western Digital would be happy to license any such patents if necessary. And if there turned out to be anything else that was proprietary to IBM and could not be licensed, Western Digital would not copy it. He emphasized, however, that nothing in their products was copied.

Moving on, Cheifet said that PS/2 add-ins and add-ons were widely introduced at COMDEX, including some clever combinations of both. For example, Rodime Systems introducts its Double Play, a 45MB internal hard disk replacement for the PS/2 Model 50. The product was intended for OEMs and dealers and promised an access time of only 28 milliseconds, which was faster than the 20MB disk included with the Model 50. The drive also shipped with an interface card for the PS/2 Model 25.

For the 5 million PC owners not ready to trade in their machines for a PS/2, Cheifet said, Intel proposed its own solution–a 386 add-in board costing less than $1,000. Owners of 8-bit Intel 8088 machines could increase their aging PC’s performance by up to 10 times, according to Intel. The Inboard 386 PC came with 1MB of RAM and software utilities to speed up screen displays and hard disk performance. Intel said the Inboard was compatible with IBM, Tandy, and Compaq machines, and would be available in January 1988.

Peripherals Not at the Periphery

Cheifet said that while it was easy to focus on the big names at COMDEX, the show was packed with smaller players whose offerings could be just as interesting. For example, Dyna Computer of San Jose, California, was ready to meet Intel head-on with a 386 motherboard replacement for PC owners. Guy Duff, Dyna’s director of marketing and sales, explained the card plugged into an existing PC/XT system, so the input/output would still be at the same speed as the original XT, which was 4.77 Mhz. But this board offered a total conversion as opposed to an upgrade.

Switching the subject from motherboards to peripherals, Cheifet said that Taiwan-based manufacturer Microtek promoted its MSF-300G scanner as a big step forward for both desktop publishing and graphic arts. The new scanner, which would be available in February 1988, could scan continuous-tone photos while recognizing up to 256 shades of gray, about 16 times more than first-generation scanners. Microtek’s John Kozlowski said that meant a couple of things for users. First, you could take all of that gray-level information per dot and send it out to a very high-resolution output device such as a PostScript-capable laser printer and it could interpolate it for higher resolutions up to 2,540dpi–effectively publication-quality output. A second benefit was that because the scanner captured so much gray-scale information, the user could manipulate that information using sophisticated graphics editing programs.

Another area of interested desktop publishers was laser printers, Cheifet said. One of the best-known laser printer manufacturers, QMS, unveiled a color thermal-transfer printer with 300dpi resolution and a choice of 3 or 4 color ribbons. At a print rate of 1 to 2 minutes per page, the $17,000 Color Graphics 100 was no “speed demon” compared to a laser printer, but its output was lightening fast when compared to color pen plotters and inkjet printers.

For computer owners looking for more conventional laser printers, Cheifet said, Japanese manufacturer Kyocera offered programmable integrated circuit cards that could store up to 512KB of data, anything from business forms to logos and signatures. Each IC card cost about $55 and was programmed using a “burner” attached to a PC. The printer then had a slot to download the information off of the burned card. The complete hardware and software system cost about $800.

Wireless Networking?

Local area networks got a big boost at COMDEX, Cheifet said, aided by fiber optics and cellular transmission. Sun River Corporation demonstrated its 386-based fiber optic LAN, the Cigna System, which could transmit data at 32MB/second. This was several times faster than conventional LANs and could even transmit full bitmap graphics.

RayNet Corporation introduced a different approach, Cheifet noted, by eliminating cables altogether. RayLAN was a “wireless” LAN that consisted of a PC adapter board, a receiver/transmitter, and Novell-compatible software. RayLAN would cost about $2,000 and ship in January 1988.

Cheifet said that modems also got a boost from cellular technology. OmniTel Inc. unveiled a 1200-to-2400 baud “modem on a chip” for connecting laptops to cellular telephones. OmniTel president Baldev Krishan said the product used off-the-shelf microprocessors. For a 2400 baud modem there were actually two processors: one for the microcontroller and one to handle the digital signal processing.

The Year of the Laptop

Wendy Woods took over narration duties for the next segment. Continuing from the COMDEX floor, Woods said that 1988 was already being called “the year of the laptop.” At least 14 companies at COMDEX displayed portable computers. Some packed the power of yesterday’s mainframes and ran on Intel’s new 80386. In fact, GRiD Systems had a laptop that actually ran on batteries.

But the most well-known name in laptops was Japan’s Toshiba, Woods said. At 15 pounds, the Toshiba T5100 was considered the lightest and most powerful laptop available. It whipped along at a speed 20 Mhz and came with 2MB of RAM, a 40MB hard disk, but a steep price of $6,500. And the biggest question was, “Who would buy it?”

Phil Vertin, Toshiba America’s vice president of sales, answered that it came down to the limits of your imagination with respect to business applications. For instance, at Toshiba’s U.S. headquarters in Irvine, California, Vertin said they took the computer to work rather than the work to the computer. He regularly brought his own laptop to meetings so that the staff could look at spreadsheets and make decisions right in the meeting.

If that much power boggled the mind and the wallet, Woods said, United Kingdom-based discount PC maker Amstrad had relief. Amstrad had some of the lowest-priced portables demonstrated at COMDEX, ranging in price from $800 to $1,000. These XT-compatibles had several features not seen before in the portable market, according to an Amstrad representative, such as a full-sized AT-style keyboard. It also worked off of regular C batteries that provided up to eight hours of power.

Woods said that as portables were becoming more versatile, so did the software. For example, Sharp Electronics offered to put custom software directly onto ROM chips used in its portable. These so-called EPROMs were available only to Sharp’s corporate customers. Sharp spokesperson Bill Robinson told Woods that when a program was loaded from EPROM it came up on the screen very fast since it was effectively working at the same speed as RAM. The EPROMs were thus not subject to the kinds of mechanical failures possible with a floppy disk and were very economical.

Atari Touted Next Generation “Transputer”

And speaking of economics, Woods turned her attention to the Atari Corporation booth. Jack Tramiel’s company was known for its low-priced, high-value machines. Atari brought a bevy of new products to COMDEX, including a new 80386-based PC clone, a CD-ROM player for the Atari Mega series, and–most importantly–a new kind of computer called the Abaq Transputer.

Atari’s Neil Harris explained that with the Transputer technology, you could actually plug in several microprocessor chips and have them operate in parallel on one project. Each processor could have its own bank of memory, and the CPUs communicated with one another over a high-speed serial interface. This meant that a single Transputer chip was approximately 7 times faster than the chips used in the Apple Macintosh II. You could potentially have thousands of Transputer chips linked into one machine, offering tremendous desktop computing power in the office.

Woods said that Atari planned to begin shipping the Transputer in mid-1988.

What About Software?

In terms of new software offerings at COMDEX, Woods said one highlight was Lotus Agenda, a personal database that the company described as “HyperCard for the PC,” albeit without sound or pictures. Aldus Corporation also showed off its latest PC program, Snapshot, which allowed a picture taken from any video sources to be edited, placed into Aldus’ PageMaker, or made into a camera-ready halftone on a laser printer. Snapshot would cost $500 and ship in the first quarter of 1988.

Intel Looked to Gain from IBM’s Confusion

Back in the San Mateo studio, regular Chronicles contributors Tim Bajarin and Jan Lewis joined Stewart Cheifet and George Morrow. Morrow opened the COMDEX post-mortem by asking Bajarin for his thoughts on the event. Bajarin noted that at last year’s COMDEX (1986), the industry was still in the midst of a downturn and there were only about 70,000 to 80,000 people in attendance. This year, however, you could tell that the industry was thriving again, because there were more people, exhibitors, and taxicab lines than ever before.

Bajarin continued that the two key themes at this year’s COMDEX were graphics and communications. With respect to the latter, he quipped that every year at COMDEX was proclaimed “the year of the LAN.” And we were getting closer to that being a reality. He also pointed to the fact that so many new laptops debuted. He counted at least 20 that were introduced at the show.

Lewis added that the processors and mass storage on those new laptops were of special note. Morrow proclaimed there was “finally” hard disks on the portables. Bajarin pointed to the Toshiba Model 1000 featuring a 1MB RAM card that effectively did away with the need for a 20MB hard disk. He said such RAM cards could allow you to download new software anytime you wanted. Morrow lamented, however, that the Toshiba’s battery was not removable.

Cheifet shifted the discussion to laser printers. Bajarin said there were three major issues that he saw at COMDEX in this area. First, we saw the first sub-$1,000 laser printers. That was a “magic number” for the industry. And even though they were “dumb” printers only capable of printing ASCII characters, it still laid the groundwork for the future. (Lewis added this made it possible for the average person to afford their own laser printer.)

The second thing Bajarin noted was that Toshiba introduced the first PostScript-clone laser printer. He thought that would have a great impact on the market because it meant Japanese manufacturers would be able to compete with U.S. companies like Apple, Wang, and DEC, which all relied on Adobe’s PostScript. Lewis noted that Adobe had been charging very high licensing fees for PostScript. And since Adobe sold PostScript to the manufacturers, that cost was marked up 3 or 4 times for the end-user. So she expected PostScript clone printers to dramatically bring down the cost of PostScript laser printers. Morrow said that once the PostScript clones lowered the price below $1,000, we’d likely see an explosion in the marketplace similar to that of VCRs.

Referencing an item featured earlier in the program, Cheieft asked Bajarin about the Intel Inboard 386 card for PCs. Bajarin said it was an interesting product. The millions of people who currently owned 8088-based PCs had been wondering whether to buy a 286 card or throw out the machine and buy a new 386. The 386 card basically allowed those users to upgrade to the latest CPU for just $995.

Morrow asked Bajarin about the prognosis for other companies manufacturing similar replacement boards. Bajarin said it would be tough. As the major supplier of the actual microprocessors, Intel would have a nice little hold on the market for now. Few competitors would be able to match Intel’s $995 price point. Morrow countered that smaller competitors could still beat Intel by having lower overhead costs. Bajarin disagreed but Lewis thought that was possible.

Cheifet asked about the new software discussed at COMDEX, notably IBM’s OS/2 operating system. Lewis said that was the major area of confusion at the show. She chaired the OS/2 panel at COMDEX and was still confused herself. IBM and Microsoft claimed the new operating system would “conquer the world.” But software developers didn’t know what to do. And while two years ago, Fortune 500 companies said they would move in the direction of OS/2 when it came out, now they were looking to convert their existing PCs using 386 boards or some similar solution. People were simply tired of waiting to see what IBM would do. And companies like Compaq had been the major beneficiary of IBM’s delays.

IBM announced its long-promised follow-up to the original PC line–the PS/2–in April 1987. The initial reaction was largely cautious optimism. Wendy Woods’ NewsBytes quoted one PC retailer who said, “It’s a solid meat and potatoes line of computers,” meaning there was “little real sizzle” but “no obvious mistakes.” NewsBytes added that not all of the promised models were available yet, nor was the new OS/2 operating system, which was originally scheduled to launch in 1988.

Overall, IBM hedged its bets with the PS/2. There were five models announced at launch. The lower-end 25 and 30 models were basically just clones of the original PC with slightly better specs. It was the three higher models–numbered 50, 60, and 80–that featured the new Micro Channel Architecture (MCA), which was the subject of all the clone speculation on the COMDEX floor.

Chips and Technologies, one of the companies Stewart Cheifet discussed in his report, was apparently the first to actually announce a clone of the chipset used in the 50 and 60 models. At a press conference in January 1988, a C&T manager claimed their chipset was actually 60 percent faster than those used by IBM and that “even the smallest garage in Taiwan” could now assemble a PS/2 clone.

Of course, no PC manufacturers actually showed up at this press conference, as IBM was demanding steep licensing fees–as high as 5 percent of a machine’s sales price–to license its claimed patents in MCA. Kaypro Corporation announced plans to produce an IBM-licensed PS/2 clone in February 1988, but that machine never materialized. A few weeks later, in March 1988, Taiwanese manufacturer Acer debuted what was probably the first PS/2 clone to make it to market–the Acer 1030–but it was based on the Model 30, so it used the original IBM PC architecture and not MCA.

So who managed to actually clone the MCA PS/2 first? I believe the answer was Tandy. Both Tandy and Dell Computer announced clones in April 1988 based on the Chips & Technologies chipset. But Tandy actually shipped its clone of the PS/2 Model 80–dubbed the 5000MC–in July 1988. And while Dell did license the MCA-related patents from IBM, I can’t find any record of the company shipping a PS/2 clone. In any case, Tandy’s clone proved to be an expensive failure, with an initial retail price starting around $5,000, which admittedly was an improvement over IBM’s $7,000 configuration for the Model 80.

It turned out there wasn’t much of a market for PS/2 clones since the PS/2 itself failed to realize IBM’s goal of replacing the relatively open architecture of the original PC line with a proprietary standard strictly controlled by IBM’s own patents. As Jan Lewis observed in the episode, IBM’s customers were tired of waiting, both for the PS/2 and then OS/2. This wasn’t 1981, when IBM faced only token competition from small manufacturers producing S-100 machines. It was six years later, and there was now not only well-financed domestic competition such as Compaq, but also the recent entry of multiple Asian manufacturers at the lower end of the 386 market.

What the Heck Was Atari Doing?

Speaking of the other end of the market, let’s briefly talk about Jack Tramiel’s Atari Corporation. As Wendy Woods mentioned in her floor report, Atari announced a lot of products at COMDEX. But those announcements failed to yield any notable successes.

Tramiel’s biggest move of 1987 actually came about two months before COMDEX, when Atari purchased Federated Group, a small chain of electronics stores based largely in California and Texas, for $67.3 million in cash. Atari made this move for one simple reason: It needed better distribution for its products.

Neil Harris, the Atari representative who spoke in this episode, outlined a lot of Atari’s distribution woes in a letter to the editor of an Atari users group newsletter that made the rounds in late 1987:

At the time of the formation of the new Atari Corporation in the summer of 1984, the 8-bit [computer] line was not faring too well in the mass merchants. It seems that computers were neglected during the last year or so of Atari Inc. The largest companies selling the computers, such as Sears and K-Mart, had taken the position that the 8-bit Atari computers were dead, and they proceeded to close out their inventories of computers, peripherals, and software at below-cost prices.

Compounding the situation was the set of records that were inherited by the new company. According to our books, many of the big accounts owed us millions of dollars for products shipped. According to their records, though, Atari owed them millions for products returned. When two companies have many millions of dollars in discrepancies on the books, it is very difficult to do business together. In fact, the K-Mart account was finally settled this past summer, fully three years after the new company was formed.

Harris went on to say that stores simply did not want to carry Atari’s products, particularly the low-end computers, as retailers now preferred “high-end $150 game systems” from Nintendo and Sega. He insisted the Federated purchase would help remedy that, however, as these stores would now carry Atari’s full line. Of course, Federated only had stores in five states at the time.

More to the point, Federated proved not to be a viable retail partner. Atari’s general counsel at the time, Nicholas Lefevre, recalled in a 2015 interview with Kay Savetz of Antic The Atari 8-Bit Podcast, that Tramiel entered into the deal suspecting that Federated was inflating the value of its retail inventory. Lefevre said Tramiel made a “bet” with Federrated’s founder and principle shareholder, Wilfred Schwartz, on this very point–and that Schwartz would have to compensate Atari if those suspicions proved accurate. Ultimately, there was litigation between Tramiel and Schwartz, and Atari ended up dumping the Federated stores two years later in November 1989.

Of course, in 1987 Tramiel was largely praised for the Federated deal. Victor F. Zonana of the Los Angeles Times called the purchase “a master stoke that will dramatically expand Atari’s distribution for its well-regarded, inexpensive line of personal computers while enhancing Federated’s ability to compete” with competitors like Circuit City. Yet even Zonana’s largely glowing report noted that Tramiel had created many of his own distribution problems dating back to his “rough treatment of computer dealers while he was at Commodore.” And one Atari users group member warned, “Jack Tramiel always operates in a start-up mode” and there were “always too few people doing too many jobs.”

The press should have heeded the warning. Yet the buzz for Atari was strong coming out of COMDEX, particularly with respect to the announcements of the “Transputer” and a CD-ROM drive. Neither product ever made it to the market outside of a few prototypes.

The Transputer was an especially bizarre move for Atari. It was an attempt to create a low-cost professional workstation, which was about as far as you could get from 8-bit gaming machines. The first rumblings of the Transputer came around May 1987, when journalist Guy Kewney wrote in Personal Computer World that Atari would ship a “next generation” computer based on the Inmos Transputer chip and not the Intel 80386. Inmos was a United Kingdom-based semiconductor manufacturer. The transputer was a type of CPU designed to work as part of a parallel processing machine.

Kewney said Atari believed it could produce a transputer machine capable of four times the performance of its ST line, which were based on the Motorola 68000 family of processors. Over the next few months the rumors continued to swirl. One report said Atari would pair a transputer chip with a 68000 in a dual-processor machine. During the October lead-up to COMDEX, NewsBytes reported that Atari would have one machine based on a 64-bit transputer and a second that would be an “add-on box” for the Atari Mega ST, the “professional” version of the ST line.

Things didn’t get much clearer at COMDEX. Atari did show off a demo unit of a $3,000 transputer-based microcomputer it dubbed the Abaq. But there was no operating system. It was still in development by an outside contractor, United Kingdom-based Perihelion Software. And there were rumors circulating on the floor that Tramiel met with Steve Jobs to discuss a joint venture between Atari and NeXT, Inc., with regard to transputer technology. That rumor proved fruitless, but it suggested that Tramiel might have been starting to lose confidence in the project.

Indeed, in December 1987, Inmos laid off most of its staff in the United States and consolidated its operations in the United Kingdom. Delays followed at Atari. While the company continued to show off its prototype at European trade shows in early 1988, that July Atari announced shipments would be delayed until “at least 1989” due to the fact the operating system still wasn’t ready.

Around May 1989, Atari did finally ship some workstations, although the name had been changed from Abaq to the Atari Transputer Workstation 800. Only about 350 units were ever produced. But at least the operating system–dubbed Helios–was finally done.

Notes from the Random Access File