In CI No 675, we presented Compaq president Rod Canion’s discussion and analysis of IBM’s new Personal System/2, in a statement that was first presented in New York last week. Today we conclude the analysis with Canion’s comments on the software implications of the flagship product IBM has designed to carry it through into the personal computer market of the early 1990s.
IBM’s architectural advances in hardware haven’t really resulted in any significant new user benefits that couldn’t have been provided within the industry standard. Now let’s look at the software announcements. The main software announcement was OS/2, a new operating system for 286-and 386-based personal computers. When it finally becomes available sometime in 1988, OS/2 will provide capabilities that aren’t available with existing operating systems. The most important capability is running the 286 and 386 in the protected mode, which provides minicomputer-like multi-tasking capabilities.
Compaq will supply this to its customers through Microsoft’s OS/2, which is compatible with IBM’s standard version of OS/2. However, there is a serious misperception about the impact OS/2 will have on the personal computer industry. Many people assume that, as in the past with version changes of DOS, there will be an automatic mass migration to the new operating system once it becomes available. But this will not be the case. For most business users, DOS 3 and its extensions will provide all the capabilities they ever need, even on 286- and 386-based systems. The features OS/2 will provide won’t be needed, and the penalty for changing to it will be significant. The protected mode is an important capability, but it will really be useful only for a relatively narrow set of applications. It will be used most in large organisations for entirely new and specialised multi-tasking application programs – for instance, to enable a user to work on inputting data while the workstation is simultaneously interacting with the mainframe and other workstations in the network. Once OS/2 enters the market in 1988, and once large corporations are able to develop special applications around it – perhaps by 1989 – then we’ll see some important problems being solved. But the cost of OS/2 is high – more than $400 per system just for the operating system. The user also pays penalties in the amount of memory that must be added to the system for OS/2 to run, and in degradation of performance of existing programs. The result will be that only the users who really need the capabilities of OS/2 will switch to it. Everyone else will continue to be happy with DOS 3. While OS/2 will work fine on 80386-based products, it doesn’t take advantage of any of the unique advances provided by the 80386 microprocessor. These unique capabilities will be provided through extentions to DOS 3. Two key features of the 80386, beyond the 286 protected-mode multi-tasking capability, make it important to users. One is the virtual mode, which provides a different and to most users a more useful kind of multi-tasking. It enables users to run current industry-standard programs concurrently. This capability won’t be possible with OS/2. The second 80386 feature is its 32-bit architecture, which enables it to run software written with 32-bit instructions, greatly increasing the speed of programs. Here, too, extensions to DOS 3 will provide this capability this iear. But it won’t be available on OS/2. New 32-bit programs will speed performance of applications ranging from traditional business productivity software, to CAD/CAM, to communications, and even to emerging technologies like artificial intelligence.These two 80386 features are even more important for the broad market than protected-mode multi-tasking. The fact is that DOS 3, with a growing number of extensions, can take advantage of more of the advanced capabilities of the 386 than OS/2. And unlike OS/2, it will be available this year. In fact, I would predict that DOS 3 applications will continue to meet a far broader set of user needs than O