The First Operating System
CP/M was 5K and it gave you no more and no less than what an operating system should do.
–Alan Cooper, personal-computer software pioneer
The first operating system to qualify as a standard in the developing microcomputer industry actually appeared before the Altair itself. CP/M was not the result of a carefully planned project involving years of research by dozens of software specialists. Like most of the early significant programs, it originated out of one person’s initiative.
Gary Kildall
In mid-1972, Gary Kildall came across an advertisement on a bulletin board that said “MICROCOMPUTER $25.” The item advertised, the Intel 4004, was actually a microprocessor, arguably the first in the world, but it still sounded like a real bargain to Kildall. He decided to buy one.
Although many of the microcomputer companies’ founders didn’t fit the typical image of an industry leader, Gary Kildall didn’t even act as if he wanted to be in the game. While wrapping up his PhD at the University of Washington, Kildall had moved to Pacific Grove, California. He loved the scenic coastal town; its laid-back, fog-draped ambiance seemed to suit him. Kildall was soft-spoken, possessed of a disarming wit, and was most at ease in a sport shirt and jeans. He was an incurable diagram drawer. When he wanted to make a point while speaking, he would frequently fish around for chalk or a pencil. In the early 1970s, Kildall was happy in his job at the Naval Postgraduate School. He enjoyed teaching and the job left him time to program. Kildall had no particular business skills and no real desire to leave academia. He was comfortable just where he was.
Gary Kildall also liked to play with computers, and knew a lot about them, in both an academic and a practical, hands-on sense. He had been one of two people responsible for keeping the University of Washington’s Burroughs B5500 computer up and running. Later, when the university was acquiring its new CDC 6400 computer, Kildall was so well respected for his knowledge of computers that he served as the technical advisor on the purchase.
The other person responsible for keeping the B5500 running was Dick Hamlet. He and three others started a time-sharing company in Seattle that used DEC’s PDP-10 computer and some new DEC software. The idea was to allow people to log onto the PDP-10 remotely in order to tap its capabilities. Hamlet’s company was called Computer Center Corporation, or C Cubed, and for a time two teenagers named Bill Gates and Paul Allen worked there after hours searching for bugs in the DEC software.
It turned out that the $25 price on the Intel 4004 applied only to volume purchasers, and besides, a microprocessor was useless by itself; it needed to be incorporated into a computer. Kildall did buy the manual for the Intel 4004, wrote a program on the school’s mainframe to simulate the 4004, and began to write and test 4004 code to determine what he might eventually do with the “bargain basement” 4004 chip.
Kildall recalled that his father, who owned a navigation school in Seattle, had always wanted a machine that could compute navigational triangles. Kildall wrote some arithmetic programs to run on the 4004, offhandedly thinking that he might come up with something his father could use. He was just fooling around with the 4004, trying to see how far he could go and with what degree of speed and accuracy. He determined that the processor was pretty limited, but he still loved working with it. Soon thereafter, he traded some 4004 programs to Intel for a development system, a small computer built around the 4004, which was in effect one of the first true microcomputers, albeit not a commercial product.
Hooked on Micros
When Kildall visited the microcomputer division at Intel in 1972, he was surprised to see that the pioneering firm had set aside a space no larger than the average kitchen for the entire division. One of the people he met there was a clever programmer named Tom Pittman, a nonemployee who, like Kildall, had been intrigued by the 4004 and was already writing software for it. Kildall and Pittman got along well with the Intel people, and Kildall began working as a consultant for Intel on his one free day a week. In this new role, he tinkered with the 4004 for a few more months until he “nearly went crazy with it.” He then realized that he would never go back to working on large computers.
Soon Kildall was dabbling with Intel’s first 8-bit microprocessor, the 8008. He was working in the same two-level mode—that is, developing the software for a microprocessor on a minicomputer—that Gates and Allen used. Like Paul Allen, Kildall wrote programs to simulate the microprocessor on a larger machine and then used the simulated microprocessor with its simulated instruction set to write programs to run on the microcomputer. But unlike Gates and Allen, Kildall had the benefit of a development system so that he could check his work as he went along by trying it out on the system.
CP/M
In just a few months, Kildall created a language called PL/M, inspired by a mainframe language called PL/I that was significantly more elaborate than BASIC. Kildall set up the development system in the back of his classroom, in effect creating the Naval Postgraduate School’s first microcomputer lab. Curious students would wander back there after class and tinker with the system for hours. When Intel upgraded the Intellec-8 from an 8008 processor to an 8080 and supplied Kildall with a display monitor and high-speed paper-tape reader, the professor and his students had a system comparable to the early Altair before the Altair was even conceived.
Kildall realized, however, that he was still missing an essential ingredient of a successful computer system—an efficient storage device. Two common storage devices on large computers at the time were paper-tape readers and magnetic disk drives. Considering how slow the microprocessor was, paper-tape storage was simply too cumbersome and expensive. Kildall set out to obtain a disk drive and did a little programming in exchange for a drive from Shugart. There was a catch: in order for the disk drive to work, a special controller was needed—a circuit board to handle the complicated task of making the computer communicate with the disk drive.
Kildall attempted to design such a controller several times. He also tried to create an interface that would allow his system to connect to a cassette recorder. But he found that he needed more than just programming talent to solve the complex engineering problem of interfacing the two machines. The project failed, and Kildall decided he was totally inept at building hardware. Nevertheless, he had demonstrated a lot of vision. It would be years before disk drives came into common use on microcomputers. Finally, at the end of 1973, Kildall approached John Torode, a friend of his from the University of Washington who would later found his own microcomputer company. “We’ve got a really good thing going here if we can just get this drive working,” Kildall told his friend. Torode got the drive to work.
Meanwhile, Kildall polished the software. At one point in late 1973, during his months of frustration with the disk drive, Kildall had taken a few weeks to write a simple operating system in his PL/M language. He called it CP/M, short for Control Program for Microcomputers. CP/M underwent further development, although it already provided the software needed for storing information on disks.
Some of CP/M’s enhancements arose under curious conditions. While he continued teaching, Kildall became involved in a project with Ben Cooper, the hardware designer in San Francisco who had worked with George Morrow on disk systems and had later started his own computer company, Micromation. Cooper thought that he could build a commercially successful machine to chart horoscopes, and he enlisted Kildall’s help in the project. The two had no interest or belief in astrology, and in fact considered it patent nonsense, but Cooper had ideas about the hardware and Kildall wanted to do the math that calculated star positions. They also figured that the result might be a commercial success. So Cooper built and Kildall programmed, and they came up with their “astrology machine,” which would stand in grocery stores eating quarters like an arcade game and printing out horoscopes. Kildall thought the machine was just beautiful.
Commercially, however, the astrology machine was a failure. Its makers placed machines in various locations around San Francisco, but the fancy knobs and dials that excited the two hobbyists irritated users—and with good reason. Customers would drop their quarters in and the paper would jam up. Kildall and Cooper were stumped on how to fix the problem. “It was a total bust,” Kildall later said.
Despite the disappointing results, the astrology machine gave Kildall his first commercial test of portions of his CP/M. In the process of programming the astrology machine, he rewrote the debugger and the assembler, two tools for creating software, and began work on the editor. These constituted the essential tools for developing programs. In addition, he created a BASIC interpreter that allowed him to write programs for the astrology machine. Some of the tricks he learned in developing the BASIC he would later pass on to his pupil, Gordon Eubanks.
As they worked on interfacing the disk drive, Kildall and Torode exchanged their views about the potential applications of microprocessors without saying much about microcomputers. They continued to believe, along with the designers at Intel, that the microprocessor would wind up in things like kitchen blenders and automotive carburetors. They thought of providing a combined hardware- and software-development system to encourage alternative uses of microprocessors. Kildall’s belief in the future of such “embedded applications” of microprocessors was undoubtedly fostered by his colleagues at Intel. At one point, Kildall and a few other programmers had written a simple game program using the 4004 microprocessor. When they approached Intel chief Robert Noyce with the suggestion that he market the game, Noyce vetoed it. He was convinced that the future of the microprocessor lay elsewhere. “It’s in watches,” he told them.
Digital Research
So Torode and Kildall, without forming an actual company, sold their hardware and software together—not as a microcomputer, but as a development system. And when Kildall, encouraged by his wife Dorothy, finally incorporated and began to offer CP/M for sale, he had no idea how valuable a program he had written. But how could he know? There were few microcomputer-software developers around.
At first, the Kildalls called their company Intergalactic Digital Research. The name was quickly shortened to Digital Research, and Dorothy, who was by this time running the company, began using her maiden name McEwen because she didn’t want customers thinking of her as “just Gary’s wife.” Digital Research’s earliest customers grabbed some stunning bargains. For instance, Thomas Lafleur, who helped found an early microcomputer company called GNAT Computers, made one of the first corporate purchases of CP/M. For $90 he gained the right to use CP/M as the operating system for any product his company developed. Within a year, a license for CP/M cost tens of thousands of dollars.
Figure 38. Digital Research Staff Tom Rolander (front row), Dorothy McEwen, and Gary Kildall (both in front of the sign) pose with the rest of the Digital Research staff in front of their Pacific Grove, CA, headquarters. (Courtesy of Tom G. O’Neal)
Dorothy later said that a 1977 contract with IMSAI was a turning point. Until then, IMSAI had been purchasing CP/M on a single-copy basis. Its ambitious plans to sell thousands of floppy-disk microcomputer systems prompted marketing director Seymour Rubinstein to negotiate seriously with Gary and Dorothy. He finally purchased CP/M for $25,000.
Rubinstein was convinced that he had virtually stolen the CP/M operating system from Gary, but the Kildalls’ perspective was somewhat different; the IMSAI deal made Digital Research a full-time business. After IMSAI bought CP/M, many other firms followed suit. CP/M was such a useful program that it was not until IBM introduced a microcomputer with a different operating system in 1982 that Digital Research faced any serious competition. The programmers who would provide that competition were still working at MITS in Albuquerque.