Computer Chronicles Revisited, Part 76 — The Compaq Deskpro 386, Zenith Z-386, and VP/Ix

The star of the November 1986 COMDEX show was the Intel 80386 microprocessor. As I discussed in my post about the Chronicles episode covering that event, Compaq and Zenith were among the first companies to announce their own 386-based PCs. This allowed these “clone” makers to get the jump on IBM, which was continuing to drag its feet on its own next-generation PC, the PS/2, which would not come out until August 1987.

Of course, as we head towards the end of 1986, the 386 was still a high-end chip meant for high-end systems. Neither Compaq nor Zenith’s new offerings were cheap. And there was not yet much in the way of specialized software or even operating systems that could take full advantage of the 386’s capabilities. That would be the main focus of this next Computer Chronicles episode from December 1986, which focused on the speed of the 386 while also looking at what these new machines could actually do for consumers.

Jan Lewis was back for her second consecutive episode as substitute co-host. Cheifet opened the program by showing Lewis an actual Intel 80386 microprocessor, noting “this is what all the fuss is about.” He added that there was no real operating system or software available for the new processor, yet people were waiting in line to buy computers with a 386 in it. Indeed, there had been more talk about the 386 than probably anything since the original IBM PC came out six years earlier. Was this just hype or was it a major development in personal computers? Lewis said it wasn’t just hype–the 386 was a “quantum leap” in personal computers. People were standing in line for 386 systems just for the raw speed that it offered for applications such as graphics, telecommunications, and computer-aided design. But the real promise of the chip was in the applications that would be coming down the road in the next 2 to 5 years.

Kildall Saw 386 as Next Step in Software Development

Wendy Woods presented her first remote segment, which oddly enough profiled regular Chronicles co-host Gary Kildall’s company, KnowledgeSet Corporation, based in Monterey, California. Woods said that very few hardware introductions had received as much attention at Intel’s new 386 microprocessor. The fast 32-bit chip was already available in a number of desktop machines at prices that definitely put it into the microcomputer category.

Woods said that KnowledgeSet recently purchased a 386 micro to assist their CD-ROM software developers. So far, the machine ran compilation tests about three times faster than the (286-based) IBM PC-AT and equally shined in calculation-intensive graphics displays. Kildall told Woods that the 386 machine was important in a development environment because it let them get through programming tasks a lot faster than before. Where before they would have used something like a single Digital Equipment VAX computer for program development, they could now have each engineer use a 386 machine. This also meant you didn’t need the more elaborate fire extinguisher systems or a specially air-conditioned computer room, which also cut down on the expense of the operation.

Closing the gap between minicomputers like the VAX and microcomputers was certain to affect prices, Woods said, and likely to give a boost to the development of new applications. Kildall added that the reason for this was not so much because of the 386 but rather the fact that the technology was now readily accessible to a large number of people. For example, you could conceivably build a commercial expert system using a 386.

Woods noted that in spite of all the publicity surrounding the 386, 32-bit chips were not new technology. But developers saw special promise in this new round of 32-bit 386 machines. Kildall said the interesting thing was that the 386 grew out of the IBM PC world and it wasn’t just the chip that was important but that it was now considered the next stage to move to. This also meant that the basic underlying functionality had to be improved considerably.

How Fast Could You Run a Lotus Macro?

In the first studio segment Michael Swavely, the vice president of marketing with Compaq, joined Cheifet and Lewis to discuss his company’s new machine, the Deskpro 386. (Another Compaq employee was present to assist in the demonstration. I think Cheifet said his name was “Len Parsons,” although he was never identified on-screen.) Lewis opened by asking Swavely who was actually buying the Deskpro and what they were using them for. Swavely said the Deskpro had been selling well among a whole range of customers. Most of these customers were looking for the power and performance that the 386 brought to their applications, such as very large spreadsheets, large databases, as well as more specialized applications such as computer-aided design, network file services, or even small multi-user systems running the XENIX operating system.

Cheifet asked Swavely to demonstrate the Deskpro 386’s speed by running a “very complex macro” on a spreadsheet. Swavely explained this was a 5,000-cell spreadsheet running in Lotus 1-2-3. It took the 386 about 16 seconds to run the entire macro, which involved recalculating and refilling the cells in the spreadsheet.

For comparison, Cheifet asked Parsons to set up and run the exact same macro with the Deskpro in 8088-mode, which restricted the machine’s clock speed to 4.77 MHz, the same as the original IBM PC. (The 386 in the Deskpro ran at 16 MHz at full speed.) While that ran, Lewis asked if the speed of the 386 rendered the AT and the 286 obsolete. Swavely said he didn’t think so. There was still a significant price difference for the added performance of the 386. So both the 286 and 386 products would have a place in the market for several years.

Cheifet asked about compatibility. Would all existing software run on the 386? Swavely said yes, Compaq was staking its reputation on the compatibility of the Deskpro 386.

Cheifet noted that some users were worried that “any day” now IBM would announce its own 386 machine and it was gutsy for Compaq to go first. How risky a move was it for Compaq? Swavely said he didn’t think it was a risk at all. Compaq believed the industry standard that had been established for business software was clearly in place. The Deskpro 386 innovated within that existing standard as opposed to trying to go outside the standard and do something different. IBM may or may not enter the 386 marketplace in the future and the market would judge them the same as any other manufacturer.

Cheifet followed up, clarifying that Compaq wasn’t worried about IBM creating a slightly different standard. Swavely said not particularly, because IBM had to live within market realities. And that reality was that American business had made an enormous investment in the industry standard. Cheifet quipped that IBM may have “buried itself” in that standard.

Meanwhile, the demonstration macro was still running in 8088 mode. As the waiting continued, Cheifet asked when we would actually start to see new applications for the 386. Swavely said the Deskpro 386’s launch had kickstarted new development, and right now there were hundreds of companies in the industry working on new hardware and software capabilities seeking to take full advantage of the 386 architecture.

As Swavely continued his response, the Deskpro beeped to indicate it had finally finished running the macro in the slower mode. Cheifet said the final time was 2 minutes and 17 seconds, as compared to the 16 seconds it took the machine in full 386 mode.

Sun Betting on Good NeWS

Wendy Woods presented her second remote report from Sun Microsystems. (This was Wendy’s second visit to Sun, having previously profiled the company in a February 1985 episode focusing on UNIX.) She opened on an image of several programs running at once on a 386-based PC. Woods said this was accomplished using a programming tool called NeWS from Sun. Previously, this product was only available on Sun workstations.

So why was NeWS news? Woods said that Sun was betting the 386 would become as pervasive at the chip that runs today. By getting to market first with a windowing product that could be adapted to many machines, Sun hoped to make NeWS the new windowing standard for UNIX. Dr. Eric Schmidt of Sun told Woods that if you looked at areas like computer-aided design and publishing, the 386 would be a major player in those markets. And he believed that NeWS would be a key component. In addition, customers who had Sun machines could run the same applications on a 386 machine using NeWS. So that also made it attractive for Sun to license NeWS to the 386 community.

Woods said that NeWS was also unique in that it was partially constructed with PostScript, a sophisticated imaging model that enabled a user to manipulate the size and scale of each window. She added that while time would tell if the 386 would be successful, there were at least two dozen software companies–many of them big names–using NeWS to write their own 386 applications.

Zenith Not Worried About IBM’s Next Move

Bob Dilworth, the president of Zenith Data Systems, joined Cheifet and Lewis for the next studio segment. Dilworth was there to talk about his company’s new offering, the Zenith Z-386. (There was once again a silent system operator to assist with the demo, but I could not make out his name.)

Lewis asked Dilworth how the Zenith machine differed from the Compaq Deskpro 386? Dilworth said the machines had similar features and both companies had to design without knowing what the other was doing. Obviously, both companies were interested in trying to protect standards. But he said Zenith’s approach was unique with respect to handling memory. But the Deskpro and Z-386 were both very fast MS-DOS machines.

Cheifet asked if the Z-386 was out now. Dilworth said it was. Zenith was shipping systems to existing customers. Lewis then repeated her earlier question to Swavely, asking Dilworth who was buying these machines and what they were using them for. Dilworth said that generally speaking it was people who needed the speed or lots of memory access, such as users handling large database files.

Cheifet asked if there was a real market for the 386 based on speed alone or would it depend on new, more sophisticated applications to move machines like the Z-386. Dilworth said that obviously new applications would drive the market. But there were currently sophisticated applications running sluggishly on even a 286 machine. So those users still needed a 386 to run their existing software more quickly.

A demonstration of the Z-386 followed. The operator loaded a 10-page WordStar document that was previously prepared by the Chronicles staff. Cheifet explained the document contained the same error 90 times. He asked to see how fast it took to perform a search-and-replace to correct those errors. Dilworth commented this operation was about three to four times faster on the Z-386 versus a 286 machine.

Cheifet again raised the question of what IBM would do. Dilworth said he wasn’t sure. He read the same press as everyone else. He expected IBM to eventually use the 80386 in some fashion, but he didn’t know when.

Cheifet asked about the consequences for users when new 386-specific applications came along. What would this actually mean for business users? Dilworth said that it was software–not hardware–that drove the market. In this particular case, he expected to see more easy-to-use software packages that required the power and speed of a 386 to work.

Running UNIX and DOS Together on a 386

Dana Krelle and Neil Colvin joined Cheifet and Lewis for the final segment. Krelle was the marketing manager for the 80386 microprocessor at Intel. Colvin was chairman and CEO of Phoenix Technologies.

Lewis opened by asking Krelle for some background on the 386 and “why all the fuss?” Krelle said the 386 was the 32-bit generation of the 8086 technology, which was also the basis for the 8088 and 80286. The 80386 not only offered compatibility with those previous-generation processors, it also represented a performance increment over the prior machines. The 386 also offered a richer architecture for software developers.

Cheifet asked about operating systems for the 386. Krelle said that UNIX System V, Release 3, had already been ported to the 386 and was now in beta at over 50 different original equipment manufacturers (OEMs) who were developing products. Krelle added that AT&T, which owned UNIX, should begin certifying some of these systems by early 1987. Additionally, several OEMs were developing products that surrounded the 386 and UNIX. A number of companies, including Phoenix Technologies, had also built tools to use MS-DOS on top of UNIX on the 386. He also pointed to Sun Microsystems’ NeWS. So at least in the workstation environment, the software market for the 386 was already very well developed.

Cheifet turned to Colvin and asked about his company’s product VP/ix. Colvin picked up on Krelle’s earlier point and explained that VP/ix was an integration of the UNIX environment for the 386 and the PC-DOS (i.e., MS-DOS) environment used on existing systems. It allowed you to run any PC task under UNIX as a UNIX task.

What was the advantage of doing that, Cheifet asked. Colvin said it gave you the best of both worlds. You could run your UNIX jobs, such as computer-aided design, and then at the same time switch to running popular DOS applications like Lotus 1-2-3 or dBASE III.

Cheifet asked Colvin to demonstrate VP/ix. Colvin explained there were two terminals on the desk to illustrate the multi user capabilities of UNIX. The terminals were linked to a 386 machine over an RS-232 serial connection. Colvin ran a standard UNIX command on one terminal. On the other terminal, he started a “virtual PC” using VP/ix. He said you could also “nest” these environments, such as running Microsoft Windows on the virtual PC, which he did. He then started a second DOS virtual machine on the other terminal to run 1-2-3.

Turning back to Krelle, Cheifet asked about the different “modes” used with the 386, such as virtual and protected modes. Krelle said there was only one mode in the processor that everyone was using, the protected mode. The ability to run a virtual machine within the 386 was all contained within the protected mode. The “protected mode,” in turn, referred to the full 32-bit architecture of the machine. Lewis clarified this meant that an application could not “clobber” the operating system. Krelle said that was correct. One reason that Intel called it “protected mode” was that there were several intrinsic features in the 80386 chip that allowed protection not only of the operating system from an application, but also from one application hurting another application, and even protection within an application so it couldn’t do “wrong things” to a machine.

Cheifet asked about the future of the 386. What would it mean for users once new applications were available? Krelle said it would mean that the applications available over the next couple of years would be much more powerful. He expected many of those new applications would incorporate artificial intelligence techniques to allow the machine to be molded to the user rather than vice versa.

AI Stocks Were Down as 1986 Closed

Stewart Cheifet presented this week’s “Random Access,” which was recorded in December 1986.

  • The year-end IBM rumors included reports that Big Blue would come out with a new low-end PC aimed at the education market, a new “souped up” 286 machine, and a new 386-based computer. Meanwhile, Cheifet said IBM had called it quits on the original PC, with sources claiming that production of the machine had stopped, despite the fact it still accounted for 7 percent of all personal computer sales.
  • Despite predictions that 1986 would be the “Year of Linux on the Desktop Artificial Intelligence,” the stock prices of AI companies fell to a three-year low at the end of the year, down an average of 26 percent from a year ago. Cheifet noted that one big factor in the future of AI applications would be the role of the U.S. Department of Defense, which up until now had been a major supporter of artificial intelligence research, particularly to support the Strategic Defense Initiative.
  • A study published by New Jersey’s Rider College found that students who used computers and word processors received better grades on papers than did students on typewriters. Cheifet said that of the 200 papers reviewed for the study, no student who used a word processor failed and 76 percent of the papers done on computers were women.
  • Montana’s legislature announced a new online database that would list all pending bills, the names of sponsors, the schedules for committee hearings, and the status of bills. Cheifet said it would cost users a $100 flat fee for unlimited access to the system.
  • Lee Foster, a California-based travel writer, came out with what Cheifet said might be the first complete book distributed exclusively on a floppy disk. Foster’s West Coast Travel California sold for $19.95.
  • A computer researcher at Bell Labs analyzed Leonardo DaVinci’s famous Mona Lisa painting and a sketch of the artist himself and concluded that based on a comparison of the facial structure, DaVinci was the model for the painting.

Compaq’s Deskpro an Early Platform for UNIX Expansion in the PC World

It’s interesting that neither Stewart Cheifet nor Jan Lewis asked about the prices of the 386 machines demonstrated in this episode. This likely belied the fact these were not computers for the typical consumer tuning into Computer Chronicles on a Saturday afternoon. Indeed, as Wendy Woods’ remote segments made clear, the 386 was initially seen by many in the industry as a means of replacing minicomputers with equally (or more) powerful individual workstations that typically relied on UNIX rather than MS-DOS.

So exactly how much was a 386? Well, according to an extensive review in the November 25, 1986, issue of PC Magazine, Compaq’s base model for the Deskpro 386, the Model 40, started at $6,499. This configuration included 1 MB of RAM, a 1.2 MB floppy disk drive, and a 40 MB hard disk drive. Some optional extras included a second 40 MB hard disk for $2,199, a 4 MB RAM expansion board for $2,999, and an Intel 80387 math coprocessor for $349. (The Zenith Z-386 base model had an identical configuration and, I’m guessing, a similar price.)

The Deskpro was designed to operate under MS-DOS 3.1 and Xenix. Xenix was a UNIX-based operating system that Microsoft originally developed under a license from AT&T. Microsoft did not sell Xenix directly to consumers but instead subcontracted most of the development and support to the Santa Cruz Operation (SCO), which later took over the Xenix project outright. PC Magazine reported in November 1986 that SCO had already shipped a development toolkit for a 386 port of Xenix, which ended up being released in 1987. SCO also distributed VP/ix, which cost $495 for a one- or two-user license or $995 for three or more users.

In the Chronicles episode, Neil Colvin presented VP/ix on behalf of his company, Phoenix Technologies. Phoenix actually marketed the software as part of a joint venture with Interactive Systems Corporation (ISC). Founded in 1977, ISC was one of the first third-party vendors to obtain a commercial UNIX license from AT&T. ISC created its own UNIX variant, PC/IX, which competed against Xenix. In 1984, IBM decided to start selling PC/IX directly to users of the PC/XT, possibly in an effort to minimize Microsoft’s growing dominance of the operating system market.

ISC subsequently released 386/IX–with VP/ix as an add-on module–to take advantage of the new Intel 386 architecture. In March 1988, Eastman Kodak purchased ISC and continued distributing 386/ix and VP/ix under its new Interactive Systems Products Division. But three years later, in September 1991, Kodak decided to sell the entire division to Sun Microsystems, which merged it into its existing SunSoft subsidiary. Sun released a final version of what was now known as Interactive UNIX in 1998.

Colvin Rose from the Ashes of Phoenix’s Stock Price to Rebuild Local Theater Company

Unlike ISC, Phoenix Technologies is still in business. Founded by Neil Colvin in 1979, Phoenix made its name by reverse engineering the BIOS software used in the original IBM PC and selling it to clonemakers. Phoenix carefully documented its “clean room” process to protect against a possible copyright infringement lawsuit from IBM, according to a July 1984 report by James Langdell for PC Magazine:

One group at Phoenix examined the BIOS software documents in IBM’s Technical Reference manual, and wrote a set of specifications that described how the program functioned without including any examples of actual code. This description was passed on to a single programmer who tried to write code that fulfilled its specifications. Whenever Phoenix found parts of this new BIOS that didn’t work like IBM, the isolated programmer would be given written descriptions of the problems, but not any coded solutions that might have hinted at IBM’s original version of the software.

Boosted by this and other early successes, Colvin took Phoenix public in 1988. But as Phoenix’s post-IPO stock price plummeted and financial losses mounted, the company went through some UK-style management turnover. Covlin stepped aside as CEO in mid-1988 but remained the company’s chairman. Lance Hansche, the company’s executive vice president, then took over as CEO for six weeks until he was replaced by Ted Joseph as acting president and CEO. More than a year later, in December 1989, Phoenix announced Ronald Fisher as its new president and CEO. Ironically, Fisher had held those same posts at the aforementioned Interactive Systems Corporation. (Fisher told the New York Times he was looking forward to moving back to the east coast, as Phoenix was based in Massachusetts.)

Ten days after Fisher officially took the reins, Colvin announced he was stepping down as Phoenix’s chairman. Colvin told the Boston Globe that he hoped “to use the chairman’s title as an incentive to woo high-caliber candidates for the board.” Although Colvin indicated he would still remain with Phoenix in some capacity, there’s not much record of his activities in the tech industry after 1990.

In 1993, Colvin funded the rehabilitation of the Orpheum Theater in Foxborough, Massachusetts. The Globe reported at the time that Colvin had long been active in promoting live theater, and he was then serving as president of the Foxborough Regional Center for the Performing Arts. The Center and the Orpheum remain active today, having renamed itself the Marilyn Rodman Performing Arts Center in 2017.

As for Phoenix Technologies, it regained its footing under Fisher and his successors and continued to primarily develop and sell BIOS software and related computer firmware. In August 2010, Marlin Equity Partners of Los Angeles acquired Phoenix’s outstanding stock for $139 million and took the company private. In January 2021, a newly formed group called Sentara Limited and led by Gerard Moore purchased Phoenix Technologies in what was described as a “friendly acquisition.” Moore is currently Phoenix’s CEO and Sentara’s major shareholder.

Notes from the Random Access File

  • This episode is available at the Internet Archive and has an original broadcast date of December 23, 1986. It was likely recorded on December 6, 1986.
  • Michael Swavely joined Compaq a few months after its 1982 founding. He served as vice president of marketing when this episode first aired in 1986. Compaq promoted Swavely to president of North American operations in May 1989. He announced a six-month “personal sabbatical” from Compaq in January 1991, which turned out to be permanent. The company announced Swavely’s retirement on July 11, 1991. Ten years later, in 2001, Swavely and another former Compaq executive, Gary Stimac, purchased Texas-based server company RocketLogix and renamed it RLX Technologies. With Swavely as CEO, RLX initially marketed low-cost servers to web-hosting companies. But the dot-com bust put an abrupt end to that plan. After burning through $30 million in funding, new management took over from Swavely and kept RLX going until November 2005, when Hewlett-Packard purchased the company from two venture capital firms.
  • Eric Schmidt is best known today as the former CEO and later executive chairman at Google (now Alphabet, Inc.) between 2001 and 2020. Schmidt earned his PhD in electrical engineering and computer science from the University of California, Berkeley, in 1982, and joined Sun Microsystems the following year. During his 14-year stint with Sun, Schmidt rose to become president of the company’s Sun Technology Enterprises subsidiary. In 1997, Schmidt left Sun to become chairman and CEO of Novell, Inc., a post he held until joining Google. While at Google, Schmidt participated in an illegal scheme with then-Apple CEO Steve Jobs and several other tech companies not to recruit each other’s employees. (Schmidt also served on Apple’s board for several years.) After retiring from Google, Schmidt and his wife started Schmidt Futures, which funds a variety of research grants.
  • Robert Dilworth died in September 2017 due to pancreatic cancer. Born in 1941, Dilworth was an executive with Varian Associates, Sperry Corporation, and Ultra Magnetics before joining George Morrow’s company, Morrow Designs, as president in 1982. Dilworth abruptly left Morrow in July 1985 to become president of Zenith Data Systems. As I discussed in my prior post about Morrow Designs, the Dilworth-led Zenith won a key contract to provide laptop computers to the Internal Revenue Services based on a design licensed from Morrow. Morrow Designs received no per-unit royalties from the deal, however, and after that company filed for bankruptcy in March 1986, Dilworth’s Zenith subsequently acquired all of Morrow Designs’ intellectual property rights. Dilworth left Zenith in August 1987 to join a startup company, Metricom, Inc., which developed one of the first wireless internet access services, Ricochet. Microsoft co-founder Paul Allen acquired control of Metricom in 1997 through his venture capital fund and Dilworth apparently retired sometime shortly afterwards.
  • Dana Krelle worked at Intel from 1983 to 1992. He then became vice president of marketing with a startup server company, NextGen, which Intel rival AMD acquired in 1996. Krelle remained with AMD until 1999. After a couple of more stops as a marketing executive, Krelle served as CEO of QuTel, Inc., from 2015 to 2020. QuTel was an apparently unsuccessful startup that tried to market intellectual property based on quantum computing technology.
  • Thomas Simonet was the communications professor behind the Rider College (now Rider University) survey. Simonet, who retired in 2017 after 43 years at Rider, surveyed 200 student papers as Stewart Cheifet mentioned. And while the computer-written papers did average higher grades, Simonet and his fellow researchers said that could be due to factors other than the computer itself, such as the lack of distractions in computer labs and the simple fact that “good students may be more likely to use computers.”
  • Lillian F. Schwartz is the computer artist who performed the analysis of the Mona Lisa. She produced a short film in 1984 on her theory, where she postulated that Leonardo Da Vinci started with a female model and made changes to the facial features to match his own. But Milton Esterow, the former editor and publisher of ARTnews, wrote in a 2019 article for The Atlantic that Schwartz’s theory was “not one that experts have rushed to embrace” over the years.
  • There was at least one 386-specific operating system available when this Chronicles episode aired. I found an ad in the November 1986 PC Magazine for PC-MOS/386, an operating system made by The Software Link, Inc., touted as “specifically designed to take full advantage of the 80386’s 32-bit multi-user power.” The source code for PC-MOS was later released under an open source license.
  • Neil Colvin will appear on Computer Chronicles again, participating in at least two of the annual “Computer Bowl” specials that aired during the 1990s.