GKTMA.WS4 --------- selections from "They Made America", 2nd Edition, Harold Evans Back Bay Books, 2006 Notes from Herb Johnson, 2009 ----------------------------- The text from the chapter on Gary Kildall was retyped by Emmanuel Roche. He added a number of editorial comments, not all of which I've included in this exerpt. Comments by Roche are entirely his responsibility. I do not agree with all his comments, so I add my own on occasion. All comments are in []'s and identifed as his or mine. Evans' apparent source for Kildall's quotes, is an autobiography written by Kildall late in his life called "Computer Connections". Therefore the quotes and their context cannot be verified. I recommend reading other publications of Kildall to date and verify those comments. This selection from the book only covers Kildall's early CP/M work. It is a violation of copyright to provide a copy of the full chapter without permission. This exerpt is made available for purposes of scholarship and historic review. - Herb Johnson] Gary Kildall (1942-1994) ------------ He saw the future and made it work. He was the true founder of the Personal Computer revolution and the father of PC software.... In the early 1970s, he was utterly brilliant at programming -- but that is an understatement of his crucial role in the Personal Computer revolution. He was the first person to realize that Intel's microprocessors could be used to build not just desk calculators, microwave ovens, traffic systems and digital watches, but small Personal Computers with an unimaginable multiplicity of uses. Then, entirely out of his own head, without the backing of a research lab or anyone, he wrote the first language for a microcomputer Operating System and the first floppy Disk Operating System before there was even a microcomputer, months before there was an Apple, years before IBM launched a Personal Computer. [ROCHE> He was PAID by Intel to develop PL/M... And, to use PL/M, he needed an Operating System.] [Herb: No. PL/M was first written in FORTRAN as a cross compiler for the 8008. Intel sold it as such in 1973.] Kildall did it, moreover, in such a manner that programmers were no longer restricted by compatibility with the computer's hardware. In Kildall's system, anybody's application could run on anybody else's computer. It was the genesis of the whole third-party software industry. This alone would have been an astounding achievement. Yet, Kildall's accomplishment, while revered by experts -- "the world changed dramatically because of him" (Dr. Ken Hoganson of Kennesaw State Universty) -- is relatively unknown to the millions of users of the PC. Professor Sol Libes summed it up: "Every PC owner owes Gary a debt of gratitude. Bill Gates and Microsoft owe him more than anyone else." At the end of his life, Kildall wrote an autobiography, "Computer Connections", which has never been published. It is incisive, unaffected, moving and funny, suffused by Kildall's romance with technology. It informs part of the narrative that follows, and is the source of the Kildall quotations, but nothing may ever be enough to drive a stake through the heart of the appealing myth of how Kildall missed becoming the richest man in the world. [Note by Herb - this references the history of Kildall, Gates and IBM over the IBM PC. Read the book for that material, it is not included here.] In 1972, a colleague showed Kildall an ad in "Electronic Engineering Times" saying: "Intel Corporation offers a computer for $25." Actually, it was offering the four-bit computer chip [ROCHE> The Intel 4004.], containing 2,300 transistors but measuring only approximately 0.8 by 0.3 inches. It was designed by Intel's young Ted Hoff for a Japanese desktop calculator, but released for general sale at Hoff's urging. The cost was $25 only if you bought 10,000 of them; the price jumped to between $45 and $60 if you bought just one. But customers using the Intel 4004 chip would first need to design a custom board-level or box-level system with memory, power supplies, keyboard, display and cables. Kildall was intrigued. He had never heard of this "little chip company", but he sent for specifications for the first development system for the Intel 4004. It was a little foot-square blue box called the SIM 4-01, with read-only memory (ROM), but the price was $1,000 plus $700 for a Teletype. He did not have enough money for both on his $20,000-a-year salary. He got around this by faking the operation of the little Intel 4004 on the big IBM 370 [ROCHE> of the NPS]. As he programmed the simulator, the limitations of the chip drove him crazy, but he saw the potential of escaping from large immobile computers. "This [Intel 4004] was a very primitive computer by anyone's standard, but it foretold the possibility of one's own Personal Computer that need not be shared by anyone else. It may be hard to believe, but this little processor started the whole damn industry... There, in 1972, my dad's navigation 'crank' had arrived in the Intel 4004, but there appeared to be some major programming work to get the crank to actually work." The Intel 4004 had no trigonometric functions, so Kildall spent months programming the chip to find Sines and Cosines. After debugging the program on his simulator, Kildall called a friend at Intel and offered to swap the Intel 4004 simulator for a development system built around a real chip, a $1,000 SIM 4-01. The Intel engineer was less interested in the simulator than in the trigonometric functions Kildall had written. They made the trade, and Kildall had his own 4004. There was a tedious year's journey to make anything of a machine that could fed data only four bits at a time, and had no monitor. Kildall describes the process: shining a Ultra-Violet light through a quartz window for 30 minutes to erase 256 bytes of space on the EPROM (erasable programmable read-only memory), so there was room for his own little program; feeding paper tape into a Teletype and then, line by line, typing a program written in hexadecimal code (ROCHE>??? This must be a misunderstanding. Kildall was not PHYSICALLY typing the hex code: the hex code was read from the paper tape (output by the compiler) and written in the EPROM.], known as machine language; fixing the typing errors by going back to the beginning; running the corrected code to load each EPROM. "We, pioneers, had to do all this stuff two decades ago, so you can enjoy your sweet little laptop while cruising placidly over Colorado at 37,000 feet... For reference, an average JFK to SFO flight takes about six hours. That's the time it takes to program twelve EPROMs of 256 bytes each, or a total of 3,072 bytes of memory." A laptop today does all this work in a fraction of a second. Nonetheless, Kildall built a briefcase computer -- "It may have been the first Personal Computer". He proudly showed it to Dan Davis, then took it around for demos, lugging with him the 60-pound Teletype. He inspired hundreds, one of them a young engineering graduate at the University of Washington, Tom Rolander, who later became important in his life. Intel, too, was impressed by Kildall's bubbling imagination, and engaged him as a part-time consultant, initially to build a simulator for the new microprocessor the company was working on [ROCHE> The Intel 8008.], which was to be more sophisticated than the Intel 4004, and ten times faster. Software applications were a low priority then at Intel; the "software group" Kildall joined part-time was only two people tucked away in a space the size of a small kitchen. Kildall devised a video game for his briefcase computer based on a 1972 idea -- something like the future "Star Wars" -- by Intel engineer Stan Mazor, a co-developer of the microprocessor. The pair of them showed it off to one of the founders of Intel, Bob Noyce, a gentle, smilling presence who occasionally walked encouragingly through the little software corner in his white lab coat. Kildall writes: "Noyce peered at the LEDs blinking away on my 4004. He looked at Stan and me, and said, bluntly, that the future is in digital watches, not in computer games." Intel had just bought Microma, one of the first digital watch companies which was, not long afterward, beaten into the ground by a flood of Japanese digital watches. Intel thus passed up an opportunity to lead the video game industry. Kildall, in a judgment that would reverberate for him, too, writes of Noyce: "He, like all of us, made some decisions that are right, and some that could have made the future unfold in a different manner." What mattered to Kildall was that, in building an industry in microprocessors, "Bob treated his people with dignity". Intel was abuzz in 1973 with the triumph of the Intel 8008 chip, which doubled the power of its first microprocessor, and Kildall was drawn to spend more and more time there. After his "eyeballs gave way", he would spend the night sleeping in his Volkswagen van in the parking lot. He became a trader in an electronic bazaar, swapping his software skills for Intel's development hardware. One morning, he knocked on the door of Hank Smith, the manager of the little software group, and told him he could make a compiler for the Intel 8008 microprocessor, so that his customers would not need to go through the drag of low-level assembly language. Smith did not know what Kildall meant. Kildall showed how a compiler would enable an 8008 user to write the simple equation x = y + z instead of several lines of low-level assembly language. The manager called a customer he was courting, put the phone down and, with a big smile, uttered three words of great significance for the development of the Personal Computer: "Go for it!" The new program [ROCHE> The simulator was called "Interp".], which Kildall called PL/M, or Programming Language for Microcomputers, was immensely fruitful. Intel adopted it, and Kildall used it to write his own microprocessor applications, such as Operating Systems and utility programs. It was the instrument for developing the PL/I-80 compiler that he worked on with Dan Davis for three years. "Gary was very visual", Davis told me. "He would design things more or less graphically, and then transfer his design into code. He even had an aesthetic about his drawings. He was very thorough, patient and persistent in ensuring his solutions were not only correct, but elegant." Kildall's reward was Intel's small new computer system, the Intellec-8. [ROCHE> There is an obvious error in dates, here. He got his Intellec-8 for PL/M, not PL/I-80.][Herb: PL/I-80 was sold by Kildall's company DIgital Research, years after using PL/M to write CP/M and other software.] It must have been the first commercial Personal Computer, Kildall notes, though no one thought of it as that. He borrowed $1,700 to buy a printer and a video display. What irritated him was that he could not operate the Intellec-8 independently of the expensive DEC PDP-10 minicomputer now installed in the Navy's classrom at Montery -- unless he could contrive a way for the Intellec-8 to store data. [ROCHE> Error: There was an IBM 370 Mainframe at NPS, and a DECsystem-10 at Intel! The S/370 was running under CP/VMS, hence the name "CP/M"...] [Herb: this does not matter, he may have used a DEC-10 via dial up or wired Teletype line.] Experiments with cassette tape did not work; then Memorex just down the street from Intel, came up with an eight-inch floppy disk for mainframes. It held 250,000 characters moved data at 10,000 characters a second (compared with ten characters a second with the Teletype paper reader) and, in theory, gave nearly instant access to any portion of the stored data without rewind or fast-forward. Wonderful -- but the communication between Kildall's small computer and the disk drive needed a controller board to handle the complex electronics, and there was no such thing. "I sat and stared at that damned diskette drive for hours on end, and played by turning the wheels by hand, trying to figure a way to make it fly. The absence of a controller for that floppy drive was the only thing between me and a self- hosted computer. It drove me nuts." The equipment sat in his office for a year, the software genius defeated by hardware. "I would just look at it every once in a while. That did not seem to work any better." [ROCHE> Kildall says he got the drive from "Shugart Associates" in his DDJ article.] He went reluctantly back to his DEC minicomputer, and built an Operating System he called CP/M, or Control Program for Microcomputers, mimicking the name PL/M. He knew the program was sound, but he still could not get it to communicate with the disk. Desperate, he called his friend from the University of Washington, John Torode, who had a Ph.D. in Electrical Engineering. Torode worked on it for a few months, and came up with a neat little microcontroller. Kildall held his breath: "We loaded my CP/M program from paper tape to the diskette, and 'booted' CP/M from the diskette, and up came the prompt: * "This may have been one of the most exciting days of my life, except, of course, when I visited Niagara Falls, one day." Kildall opened a file, stored it on the floppy, and it appeared in the directory -- common place stuff now, but a dramatic achievement then, the world's first Disk Operating System for a microcomputer. As Al Fasoldt writes, without a Disk Operating System, a computer is just too dumb to do anything useful. Walking back to Kildall's home for a celebratory bottle of wine, the programmer and the engineer told each other: "This is going to be a big thing". But where was the market? Ben Cooper, an entrepreneur from San Francisco, paid Kildall to write a program for an arcade astrology machine he was making: Put in a quarter, dial your birth date and out comes your future. Kildall built the software system in a converted tool shed behind his home. When Ben mistakenly entered the command "del *.*" instead of "dir *.*" to get his files, he deleted all of the files on the diskette. And that is the origin of the prompt: "Are you sure? (Y/N)". [Herb: I suggest checking other sources for details and confirmation, that this arcade machine actually used a floppy drive and CP/M.] Cooper finally got his machine installed on Fisherman Wharf in San Francisco, and the entrepreneur and programmer sat on a bench one Summer evening, to see what would happen. A hand-in-hand couple put in a quarter, did not bother with the dial, and walked off happily enough with someone else's horoscope. "Because of it", writes Kildall, "they are probably married with seven children to this day". But nobody wanted to buy the 200 machines Cooper had built. The first big break was a sale of a word-processing program in 1975 to Omron, which made newspaper display terminals. It was the first company to build hardware using CP/M. Kildall and Torode split the $25,000. Earlier in the year, in Albuquerque, New Mexico, Ed Roberts [of MITS] had come out with a mail-order kit for hobbyists for the first commercially successful Personal Computer, the Altair, which sold for $500. It had an Intel 8080 microprocessor inside -- a larger, faster and more capable successor to the Intel 8008 -- with indicator lights and toggle switches on the front panel for entering programs. It was notoriously difficult to use, with only 256 bytes of memory, and no screen or keyboard. A company called IMSAI in San Rafael, across the Golden Gate Bridge from Silicon Valley, had promised delivery of a Diskette Operating System for the general public, but had not even begun to figure it out when Glenn Ewing, a former naval student of Kildall's, engaged as a consultant, told IMSAI about CP/M. "Glenn came to my toolshed computer room in 1975", writes Kildall, "so we could 'adapt' CP/M to the IMSAI hardware. What this means is that I would rewrite the parts of CP/M that manage things like diskette controllers and CRTs (screens). Well, come on, I had already done this so many times that the tips of my fingers were wearing thin, so I designed a general interface, which I called the BIOS (basic input/output system) that a good programmer could change on the spot for their hardware. This little BIOS arrangement was the secret to the success of CP/M." Kildall had, in essence, created a digital pancake. The underside could be adapted to fit different hardware configurations. But the top part was truly revolutionary; it did not have to be rewritten. Kildall developed a general-purpose easily expandable mechanism through which any application program could interface with his Operating System, by executing a simple "CALL 5" instruction This was a phenomenal advance. It liberated software from hardware. Any application could, thereafter, run on any computer. Another way of visualizing the revolution comes from former DRI programmer Joe Wein: "Kildall's Application Program Interface created a virtual program 'socket'." According to Kildall, he and Ewing completed the system on a lovely afternoon, sitting in the toolshed behind Gary's house on Bayview Avenue in Pacific Grove with its hummingbird feeders, a pastoral scene for a computer revolution, for that is what it portended. (It was still there in 2005, and ought to be an historic site.) Here, they created the "universal software bus" to run programs on any home-brew computer based on the Intel 8080; a lower-cost Intel device dubbed the 8085; or on a more sophisticated microprocessor, the Zilog Z-80, prodded by an Intel spin-off. Kildall's friend and future partner, Tom Rolander, puts into context what we take for granted today: "Think how horrible it was for the software vendors before that time. They would have to have different copies of their program configured to different pieces of hardware -- and there were scores of specialized pieces of hardware. Imagine a world where each model of car required a different kind of gasoline -- that's what it was like for computer operators, before Kildall's innovation." Kildall created the bedrock and subsoil out of which the PC software industry would grow. He licensed his system to IMSAI for $25,000 and felt rich. Clearly, there was a business here, but Kildall found the transition from inventor to innovator wrenching. His happy marriage (with "two great kids, Scott and Kristin") hit a reef in 1974, but it was retrieved by Dorothy's willingness to help make a business out of the CP/M program. She had not had a formal college education, but she had worked in a phone company's customer service department and, as Kildall writes, often outsmarted the grads who came to him. Gary continued to teach at Monterey while Dorothy handled the early business, sending diskettes to customers responding to a $25 advertisement she and Gary had bought in the famous insider magazine "Dr. Dobb's Journal of Computer Calisthenics and Orthodontia" at the suggestion of its founding editor, Jim Warren. Demand for the diskettes was slow at first; the market was made up of early computer enthusiasts. "We started in a corner of the bedroom", Dorothy told us. "There was no long-term plan. We put no money into the operation. We didn't have much savings. We lived off Visa and MasterCard." [Herb: Roche questions the $25 ad in Dr. Dobbs. I do too, as Dr. Dobbs did not SELL OR PUBLISH ads when Jim Warren talked about CP/M in the magazine.] Gary had fun with his classes at Monterey, where the graduates revved up on his enthusiasm and readiness to give everyone a chance. He led them through the steps to design a wristwatch computer that monitored a Navy diver's nitrogen pressures at varying depths, to avoid the "bends". His classrom, in the words of Michael Swaine, editor at large for "Dr. Dobb's Journal", was probably the world's first academic microcomputer lab. But it was a time to move on. [Herb: Jon Titus and other adademics published books and articles about 8008 and other microprocessors before 1975. Kildall's lab was not "first".] "He just loved teaching", said Dorothy. "It was a hard decision for him to quit school full-time." But Dorothy encouraged the choice they made in 1976 to start a full-time mail-order business they called "Intergalactic Digital Research" -- "Intergalactic" only because someone else held claim to "Digital Research" for a couple of years. They moved to an office on Lighthouse Avenue in Pacific Grove, where Gary worked from a cupola on top of the building. He initially proposed selling his system for $29.95 a disk, i.e., giving it away. At Dorothy's insistence, he asked $70 -- which was still absurdly cheap. She remembers going down the block to the Pacific Grove Post Office in 1976, hoping to find checks that would keep the company alive a little while longer but, by 1978, it was a roaring success, leaving other proprietary systems in its wake. CP/M made the Intel Operating System look like a scam. In addition to being cheap, Kildall's system was small, it was fast, and it would run on all Intel computers and competing Zilog Z-80s. "No other software product had been priced our way before", Kildall writes. "OK, CP/M's price came up to $100 per copy with Version 1.4, but no one seemed to care." That denomination was, in itself, another Kildall invention: The first digit was a "major" revision, and the [ROCHE> digit following] the decimal point indicated a minor revision for update. "You charge the manufacturers and customers a 'minor' fee to get the minor revision, and then issue a 'major' revision, like CP/M 2.0, and charge a major fee. That became the way microcomputer softare was labeled, and for that purpose only." In 1978, when sales were $100,000 a month with a 57 percent profit margin, Gary and Dorothy moved the business into a more spacious old Victorian house at 801 on Lighthouse Avenue, overlooking the waves of Monterey Bay. Gary worked on the top floor and Dorothy ran the business office on the ground floor, Dorothy abandoning the name Kildall for her maiden name, McEwen, to avoid the aroma of a mom-and-pop operation. "It was a very exciting time, and we were just very naive about everything, about starting a business, about the industry", Dorothy recalls. "We were young. The grownups had not come yet." They gradually recruited a young staff, students, professors and friends, and installed the programmers out of sight on the second level of the house. The hiring was casual. Dan Davis was laid up at home, recovering from a motorcycle accident, when Kildall walked in with a computer and asked if he would like to work full-time on a language compiler. "This was a wonderful period in DRI's history", recalls Dan. "Beer and pizza every Friday. People like John Pierce, Kathy Strutynski, Dave Brown, Bob Silberstein and others working like crazy and having a lot of fun." The atmosphere was certainly zany; as Kildall put it, a lot of marriages, a bunch of babies. People came to work barefoot, in shorts and in hippie dresses; anyone in a suit was a visitor. One candidate for interview with Kildall found herself talking technology with a red-bearded Roman imperor in a toga. Tom Rolander, visiting Kildall after three years working as an engineer at Intel, remarked that, as a pilot, he recognized the model airplane on Kildall's desk. Within minutes, Kildall bundled him into a sports car for a fast drive to the airport and a flight in the real plane, a Cherokee 180. Two days later, Rolander was at work in Pacific Grove writing the multitasking version of CP/M. But everyone worked hard. Strutynski was known as "the mother of CP/M 2.2", the largest money-spinner for DRI, for the hundreds and hundreds of hours she put in with Dave Brown and others, perfecting Kildall's original design. [Herb: the chapter goes on at length about Kildall's and Digital Research's activities, through Kildall's death. again, it would be a violation of copyright, and beyond my research intentions, to include that material here. Buy or borrow the book. - Herb Johnson]