The Highs and Lows of PC CAD History

Opinion

I wrote this article for the 70th anniversary issue of Design Engineering magazine. It is reprinted here with permission.

It was in the late 1970s when CAD for personal computers emerged. The first micro-computer, as the machines were then called, was 1975’s Altair 8800, which you had to solder together, and then program by flipping toggle switches on a front panel. It cost $2,700 in today’s dollars.

The lure of the microcomputer was strong. It promised us freedom from time-shared (and cost-shared) mainframes and mini-computers. The promise unleashed a pent-up demand that encouraged entrepreneurs to quickly advance the technology, and by the late 1970s we could buy computers that looked like today’s; some even came with applications more useful than the inevitable Space Invaders game.

The unleashing, however, had its downside. There were so many brands and standards that we became wary of jumping in: we might stuck with an expensive computer that might prove to be a dead-end. My first microcomputer from 1983 was a not-quite-IBM-compatible that cost $12,000 (with dual floppy drives) in today’s dollars, and that went obsolete two years later.

The standards mess ended in August 1981 when IBM introduced its PC, short for ‘personal computer’. It was relatively expensive, but it was (somewhat) open and, because it was from IBM, the most important computer company at the time, immediately set the standard upon which we continue to operate computers today: BIOS, ISA bus, VGA graphics, Ctrl+Alt+Del, Intel CPU, DOS/Windows, and so on.

The Start of PC-based CAD

Computer-aided design software for PCs began the same way. In the late 1970s, lots of 2D CAD programs emerged from small companies, each one incompatible with the next. I think the very first one was eventually named VersaCAD. Unlike computer hardware, prices could be quite cheap, starting at $100 for a decent 2D drafting program like Generic CADD.

By the mid-1980s, CAD vendors were vying to be #1. VersaCAD had the advantage of being first. Architecturally-oriented DataCAD was endorsed by the American Institute of Architects. Mechanically-oriented CADkey offered true 3D. But it was AutoCAD that took up the torch, even through it wasn’t first, wasn’t endorsed by any organization, and most certainly wasn’t 3D. What it had was customization: buyers of the $1,000-program could make it work they way they worked in their niche industries, a concept unique for the times.

As with IBM’s somewhat-open PC, Autodesk’s somewhat-open AutoCAD meant that people could buy with confidence. By the late 1980s, the industry began its first attempts at standardizing on AutoCAD’s file formats, so that competitors could lure away customers. First with Autodesk’s open ASCII DXF format (the attempt failed), and then with the closed binary DWG format, which succeeded. Later, the building industry standardized on IFCs (industry foundation classes), a metadata exchange format devised by Autodesk with AutoCAD R13.

A Dutch company, Cyco International, was the first to figure out how to read DWG files, and so for the first time, we no longer needed to pay several thousand dollars in buying AutoCAD just to view and print drawings. The first non-Autodesk program that used DWG as its native file format was 1996’s Vdraft from tiny software firm Softsource, followed two years later by IntelliCAD, which made DWG universal. Competitors rushed to add DWG read and write capabilities to their CAD programs, whether a $20,000 Catia system from France or a $0 FreeCAD from a community of enthusiasts.

Overcoming The Big Slog

We could run CAD on microcomputers, hurrah! But it was slow. Painfully slow. At first, a single zoom or pan of the famous Nozzle drawing took five minutes. Adding Intel’s extra-cost numeric co-processor chip ($1,300 in today’s dollars) to the computer reduced the wait to a mere 30 seconds. The break-through came in 1986 with AutoCAD v2.5’s near-instant zooms and pans through software-based display-list processing and the (for the times) hyper-fast 80386 CPUs. I recall an Autodesk salesman boasting that if you looked too closely to the screen, the speed would cut your nose right off.

CAD on our little computers started off cheaply enough, compared with industry giants ComputerVision and Integraph, who sold their proprietary turn-key systems for $100,000 or more per station. But then CAD on our little computers became expensive, as vendors raised prices with each new release and as we added hardware peripherals. We deployed pen plotters to produce paper copies of our electronic drawings, and the very best one, the fastest and most accurate HP Draftmaster, cost $45,000 in today’s dollars. About half as expensive were large 19” high-resolution (1024x768) monitors with dedicated graphics boards, but even more expensive were high-speed, large-format scanners. That thousand-dollar PC CAD program could become a $100,000 system rather quickly.

A CAD manager told me how he afforded those prices. When times were good, he would buy more CAD equipment. And when times were bad, he’d lay off some drafters and then use the savings to buy more CAD equipment. The PC CAD industry to this day never felt a recession, something I can vouch for: during my 39 years writing about CAD, I experienced none of the slowdowns that took place in the general economy.

The Windows Awakening

CAD programs first ran on Unix, on DOS, or on proprietary operating systems, which limited their capabilities. Unix was user-hostile; DOS was neither multi-user nor multi-tasking (and stingy on memory); proprietary systems bled customers dry. The solution emerged from Xerox’s research lab in Palo Alto CA: the WIMP interface -- windows, icons, menus, pointer -- which was firstly popularized by Apple in LisaOS, and then by Microsoft in Windows. WIMP removed the earlier barriers, giving us user-friendly multi-user multi-tasking operating systems that handed applications near-unlimited amounts of memory with which to work.

WIMP was, however, such a new paradigm that the CAD industry at first was reluctant to embrace it, though not surprisingly so. Windows was initially CAD-hostile. I recall being flown by a CAD vendor to its headquarters to explain “how Windows works.” It took a entire decade before Solidworks 95 showed that advanced CAD could run on Windows quite well. But what made CAD plausible on Windows had nothing to do with CAD.

Early on, Microsoft chased gamers, doing what it took to make 3D interactive games run smoothly (through high frame rates) by optimizing Windows code and working closely with graphics board manufacturers. (Curiously, Apple never did this.) We in the CAD world got a free ride on the coattails of gamers, because those advances helped make our 3D interactive design software run smoothly. Today, we effortlessly engage in real-time rendering and benefit from instant drafting previews, like of hatch patterns and fillet operations.

The March of Technology Stumbles

Within a decade, Windows and Intel CPUs were victorious. But the technology industry was dissatisfied and ever on the lookout for new sources of revenues. To great excitement, it launched tablet-size computers in the mid-1990s, which flopped. Not so hand-sized PDAs, like the Palm 1000, which were hugely successful but hardly handy for CAD. The Palm’s wildly triumphant 2007 offspring, the Apple iPhone, initially sucked at tasks like CAD, but competition with Google’s Android ramped up the speeds of CPUs, the capacity of memories, and the resolutions of screens to make portable CAD possible on phones and the newly revitalized tablets.

Despite the obvious benefits -- all your CAD drawings with you at all times! -- the portable CAD market never made it big. Apple, on purpose or by accident, standardized the pricing of iPhone apps at 99 cents (minus 30% commission) on all linked devices with free lifetime upgrades. You can’t write a CAD program at 69 cents/customer.

So, techies in their Silicon Valley cubicles came up with new ways to make money: collaboration and the cloud. We could effortlessly work together on our drawings with anyone -- at the same time! --, and save our precious IP to off-site storage -- no more worrying about fires or floods destroying the contents of our office computers.

What extroverted marketing departments didn’t understand is that engineers tend to be introverts who aren’t big on the whole real-time sharing thing. One of my favorite cartoons shows a drafter in front of a big screen with lots of faces: “They were waiting for Ralph to start designing.” Collaboration remains embedded in CAD, but tends to involve viewing and marking up drawings, once nearly finished.

When the cloud was first introduced, it felt like the shortsightedness of an AutoCAD-1.0 moment all over again: “The sky’s the limit!” we were told. The then-ceo of Autodesk proclaimed in 2015 that within three years his software would be available only online. A decade later, the promise remains unrealized.

The cloud proved to offer too many drawbacks to become popular with CAD users. They didn’t trust storing their expensively-produced intellectual property on someone else’s remotely-located computers -- what with incessant stories of hackers breaking in and stealing files. Cloud-based software could not compete in features, speed, and always-available nature with CAD programs running on desktop computers. The best effort, from Onshape, spent $100 million writing a pure cloud-based mechanical CAD program that was brilliant, but in the end amassed just five thousand paying users before selling itself to PTC. Finally, there was the pain of paying over and over again (subscriptions) to access those functions online, as servers are expensive for CAD vendors to rent.

The cloud had brought us back the bad old pre-microcomputer days.

?