|
Digibarn Stories:
Lee Felsenstein viewing and Bob Frankston commenting on the
Video interviews with Bill Pentz (by Allan Lundell)
A rendition of the story (by Bill Pentz with some additional wording by Bruce Damer) In August of 2008 Bill Pentz first visited the Digibarn to give us his early IMSAI computer with considerable original software and documentation. As we interviewed Bill we realized he was involved in the not well known personal computer history that we established the Digibarn to preserve and share for future generations. Bill’s version of this early history differed so strongly from established histories that we were confused. At the same time Bill was not someone to ignore as he taught university computer science engineering for over thirty years and his hardware and software donations showed he was clearly very technically knowledgeable. We asked Bill if he could share more about that early personal computer history that predated the Altair. He shared considerable information and thought he might find the early 8008 computer that he and his team used with some early medical software applications. Bill said he gave his friend John Moorhead that equipment years ago. We were very excited when Bill not only located John, but John agreed to let Bill donate that equipment to the Digibarn.
We already knew that CTS who later became DataPoint had engineered an Intel 8008 micro controller chip into a working computer in the early seventies years before Altair announced their personal computer. CTS then sold their units as “smart” video displays and key data entry stations because they did not feel they could market their product as a real computer. It had too little storage and programming space to run “real” programs plus everyone “knew” real computers took up large rooms, needed raised flooring, monster power supplies, etc. What Bill gave us shows that Tektronix, Inc. of Beaverton Oregon was doubly involved during that same era with this same early technology. The Tektronix 4023 graphics display terminal contained its own processor to do graphics plus Bill donated the parts that show with etched in 1972 and 1973 dates that Tektronix was already using fast serial interfaces, cassette tape interface, internal modem, and various memory cards that did not appear in the personal computer world until at least 1975. Additionally, Bill shared a whole host of different things that all came together between 1972 and 1975 that we believe helped move microcomputer based computing forward. Curator Bruce Damer note: Jack Rubin's comment/correction about Datapoint (5 Jan 2009):
Bill shared a most amazing history! It turns out that Bill led a team at California State University, Sacramento on the COMERs (COmputerized MEdical REcords System) system. This system was commissioned by Garry Gordon, MD who was president of the American Medical Preventics Society. Dr. Gordon was one of the first physicians to recognize the significant damage that people built up from years of too much lead exposure. Dr. Gordon and his fellow physicians were also interested in a long term study of that approach to also reduce atherosclerotic plaque buildup. Sac State upper division Computer Science major Gary Johnson was charged with writing a host computer program that followed the Sacramento Medical Preventics Clinic patients over time to help prove the value of the heavy metal detoxification and assist physicians to find problems before they became acute. The cost to use the campus host computer was so high a few of the easier applications involved in this COMERS project were ported over to run on an Intel 8008 microprocessor system. We believe the result is one of the first ever complete computer systems run by a microprocessor. It all started in the spring of 1972 after Bill arrived at his new job with the new Computer Science Department within the School of Engineering at California State University at Sacramento aka "Sac State". Bill found a couple of boxes in the secured digital parts cabinet that contained just-shipped Intel 8008 micro controller chips. These were Intel's first step beyond their 4004 which changed the already reeling electronics world. A bit of Background: the Post-War March to VLSI The electronics world did a near total flip flop when the tube based technology of World War II was replaced by transistors. Firms that embraced the new transistor technology thrived while those stuck supporting and selling tube technology mostly soon went bankrupt. Why would anyone buy a tube based TV for $500 when $250 would buy a far better and more reliable transistor based unit that also displayed in pristine color? This cycle repeated itself again when the aerospace technology of the sixties replaced whole printed circuit boards with single integrated circuit chips. The time between tubes and transistors was about twenty years. The time between transistors and integrated circuits was about ten years. And the time between integrated circuits and the next major change with very large scale integrated (VLSI) circuits was bare five years. VLSI circuits of the late sixties and early seventies created a huge bank of almost every type of digital function onto a large single chip. Unlike prior integrated circuits where the function of the integrated circuit chips was set during manufacture, how these digital electronic function were combined on some VSLI chips was controlled by an internal read only memory (ROM). Most hear the word memory and immediately think a ROM is something that stores computer information. Although a ROM can do this, how a ROM works is more like having all prewired with fuses that will let an electrical engineer break all unneeded connections. We call this ROM burning because we actually burn out the unneeded connections. There are so many connections involved with a VLSI circuit that electrical engineers used a computer program to oversee burning away the unneeded connections. That is why setting up a ROM became known as ROM programming. For VLSI circuits where the ROM was part of the chip, this internal rewiring using a programmable ROM (PROM) programmer became known as firmware programming. Just like building a dedicated circuit the result left a single function hard wired circuit. Because VSLI chips were expensive, a number of organizations built specialized computer programs that let them test their “firmware” before they actually burned their VSLI circuits. As VLSI circuits became ever more complex many added random access memory (RAM) to the internal VLSI ROM memory. This RAM allows a circuit to change its function based on the value in each internal memory location. Testing these variable circuits caused considerable grief. The MicroData minicomputer corporation built a very powerful general purpose minicomputer system that used electrically erasable programmable read only memory (EEPROMs) that let that computer be programmed with the firmware to create just about any kind of VLSI or computer circuit. Bill Pentz was the firmware expert for Sac State and he wrote an advanced firmware test program that permitted testing these complex circuits before they were burned. It turned out that system also provided a quick way to find problems in existing circuits, so soon Sac State was involved in many different VLSI computer and complex circuit implementations. IBM gave Sac State a nice grant in trade for turning their very popular IBM System 3 into a viable design that could be made with existing VLSI microprocessor technology and only 4k of ROM. That work earned Bill a weird reputation as the guru Sac State kept locked in a tiny little room that they fed coffee and raw meat. Bill insisted the raw meat rumor was not true but he did like coffee, lox, bagels and cream cheese. Although Sac State gave away that software, many early minicomputer firms that transitioned to using VLSI technology called upon Bill to help with debugging their projects. Bill also helped with the DEC PDP 10, Varian minicomputers, and helped model other systems. In fact, in Tracy Kidders’ book “Soul of a New Machine”, Bill was one of if not the university guru that Data General went to for help when their microprocessor design stumbled. The bottom line is that with the right software support a company in a few days got more done than they used to be able to do in years of work, plus could replace a good portion of the need to make specialized boards. In fact VLSI cards with standard interfaces rapidly replaced many specialty circuit boards. It all Started with a Bet Bill, on a bet with a faculty member, used a DigiDesigner prototyping board to turn one of the Intel 8008 VLSI general purpose controller chips into a simple but working computer in mid 1972. Unfortunately, the result was so fast that none of the Sac State analog or digital test equipment could track the signals to determine why this plug in wire wrapped system would not work reliably. Frustrated with that poor reliability, Bill got help. Russell Light, the Sac State head electrical engineering technician designed a more stable circuit. Dave Mack, the other Sac State electrical engineering technician turned that circuit into a working printed circuit board. This made a much more stable computer, but the result was still useless for any real work. It still had irritating but minor reliability problems that were eventually found to be power supply problems, but the big issues were it took at least ten times longer to program to do any practical work and lacked enough storage to run more than very simple programs. Tektronix to the Rescue! Bill got a big break on the reliability and storage problems. He explained the 8008 reliability problem to Steven Heitmann, who was a friend from college who worked as a scientist for Tektronix Inc. in Beaverton Oregon. Steve wanted Bill to work for Tektronix, so Bill visited and interviewed with Tektronix. While there Steve showed off the incredible Tektronix manufacturing ability. Tektronix designed, built, and gave Bill a full 8008 system with control panel. This unit gave Bill his reliable "dream machine" that worked with the Tektronix 4032 graphics terminal. It also gave Bill quite a bit of memory in the form of 4023 memory ROM/RAM expansion cards. That 4023 was not only one of the best ASCII terminals, but also had its own intelligence allowing it to show graphics way before its time. With the help of the Tektronix 4023 expansion boards that system also doubled as a digital development workstation which had buss based logic probes on every line five years before the first simple logic probes were sold. This Tektronix system created a stable platform and was an amazing machine which may well have been the "first full microcomputer" able to talk to a hard drive. The Tektronix display that held this system could then be expanded with a cassette interface, Teletype interface, paper tape interface, printer, and a modem for serial communication with a Cyber host at up to a whopping 9600 BAUD (unfortunately the Sac State host computer could only put out 110 BAUD)! These Tektronix terminals soon got color graphics at least five years before the first personal computers supported color or graphics. Although this was an impressive hardware collection it still was not that practical because it took graduate computer engineering skills to program and took at least ten times longer to program than existing host and mini computers. Also, the very expensive memory and limited Intel 8008 memory maximum precluded all but very short programs. Bill's Team Gets Down to Programming the 8008 Fortunately, Bill had an available fairly easy solution to solve the programming time overhead. Each different type of computer uses a different set of machine code instructions. Machine code instructions are the binary instructions stored in RAM that define what electrical logical functions a central processing unit (CPU) follows when a computer program is run. Assembly languages create one binary instruction for every human readable assembly language instruction. Because IBM dominated the computer market all Sac State computer science students had to learn BAL (Basic Assembly Language) used on the IBM System/360, 370 and later mainframes. Students had to prove their mastery of BAL by writing a working assembler, BASIC interpreter, simple compiler, operating system and data base. To stop cheating completed student program listings, card decks and outputs were collected and stored in a locked area that faculty members reviewed to ensure students did not copy prior programs. Bill recognized that all he needed was for the Intel 8008 to run the same BAL instructions. Bill had just finished a large graduate student project that created the firmware to make an IBM System 3 minicomputer run on a 4K based VLSI based computer. Bill simply changed that code so BAL also ran on the Intel 8008. This ability to run BAL let Bill's team pick and choose between the best student programs. They soon had the Sac State 8008 running DOS (Disk Operating System) which allowed loading and starting programs stored on paper tape, cassette tape, cartridge tape, and even their mobile pluggable 3/2 (three megabytes fixed, two removable) hard disk system. They had it running a simple BASIC interpreter. Because the BAL firmware ran so slow, Bill’s team also built a BAL assembler which instead of putting out one machine code instruction per human readable instruction put out all the code needed to run in 8008 machine code. The Little Microprocessor that (almost) Could, Was Made to Do a Lot!
Unfortunately, the 8008 still just did not have enough memory to run very large programs (8K addressable which in Bill's implementation they created two memory banks, one protected, one writeable). Regardless, this system still permitted using the Tektronix graphics screen, communications devices (aforementioned modem), paper tape and cassette interfaces, a printer (a Teletype model 40, which was a 300 line per minute printer versus our eight to ten character per second Teletype ASR-33), and use a minicomputer standard interface to also read and write to hard disc storage. It should be noted here that no microcomputer system achieved this type of integration with peripherals and a built in assembler plus operating system until well after the hobbyist movement got going in 1975. It was not until closer to 1978 that microcomputer operating systems actually provided this same level of standardized I/O and more than just the ability to load and run programs. The speed of the 1973 Tektronix 9600 baud serial communications interface was unheard of before about 1980 when high density floppy disks and higher speed modems came on the horizon. It should also be noted that the famed Altair 8800, Homebrew Computer Club and Apple Computer were still years in the future when the Sac State machine was up and running with all of these features.
Running BAL and having hard disc drive access transformed the 8008 microprocessor from a problematic and limited device (it had serious timing problems) into a limited but working computer. The host computer costs for the COMERS medical project were so high, that Bill and his team ported over some of the simpler applications and started running them on that 8008 system in late 1973. Eventually that system was running a nutritional analysis, medical history, genetic history, and graphical output program to take advantage of the Tektronix 4023 color graphics. This system also replaced the need to work with 80 column punched cards for entering data onto the mainframe. Bill's background in writing firmware now known as microcode was an essential ingredient to the success of this project. CSUS shared most of what Bill and his team had done with Intel in hopes of trading future help for more equipment and support. Intel asks Bill for Help on the 8080 In 1973, as Intel was developing the 8080, successor to the 8008, Bill was asked by them how they could improve the 8080 based on his experience working with the 8008. His initial response "bundle our BAL/DOS system with it and you will have a real computer" or "add more addressing modes and instructions, and here they are". Intel balked at including the 1702 PROM-based BAL as these chips were famously expensive, the results ran terribly slowly and IBM owned that proprietary language. Intel did take Bill's advice on the instructions. Meanwhile these primitive VLSI general purpose controller chips finally got named microprocessors. This made sense because they had all of the main features of the simple processors found at the heart of mainframe computer central processing units, but were microscopic in size and were minimal functioned. About this time Intel went out looking for a consultant who would take the CSUS work and take their pretty box and come up with their own unique programming language to turn the new 8080 into a fully functioned computer that could compete with the low end minicomputer systems. A young consultant/software engineer named Gary Kildall was hanging around Intel and considering taking a contract with them to help productize such a complete commercial microprocessor based minicomputer system. He was given copies of all the 8008 and 8080 work done by Sac State as background for this possible project. Gary opted not to take the contract and instead went on to write CP/M (Control Program for Microcomputer) and form his company Digital Research and the rest is history. Curator Bruce Damer: I have include a note of clarification regarding Gary Kildall's role, sent by Herb Johnson (5 January 2009):
How History Could Have been Changed (by Bruce Damer) It should also be noted that if Intel had taken Bill's advice and marketed a commercial microcomputer with BAL/DOS from Sac State it is quite likely that today we might not have a Microsoft (not such a big need to create BASIC on paper tape when you have a real assembler and OS already in ROM) or even an Apple (Steve Wozniak could have just used a "Sac State machine" to write all the color games he wanted, no need to form a company with Steve Jobs when the machine already did everything you wanted). One thing that occurs to me is that in 1975 the Homebrew Club members got their hot little hands on 8080 microprocessors all ready and working for their microcomputer kit projects but never asked "hey, who debugged this, who turned the not-so-functional-beta-versions of this into the well-its-ready-to-build-stuff-with 8080?". There is always a story behind the story. Of course, like everything else, the 8080 did not emerge pure and functional "from the void". For this computer collector and curator the story of Bill Pentz and the Sac State Machine has been one of the most satisfying projects ever here at the Digibarn. News Spreads of the Sac State 8008
The 8008 was packaged as a kit and offered by Intel in November of 1973. The cover of the MCS-8 manual shown above pictures the box designed by Sac State and contains much of the experience of the Sac State team gained while working with that early 8008. Curator Bruce Damer note: Jack Rubin's comment/correction about the MCS-8 (5 Jan 2009):
When Popular Electronics announced a working Sac State designed microprocessor based computer (we are trying to find that news item, it is not in the July-December 1974 Poptronics, anyone have other issues to check?), that still infant department was deluged with phone calls, many from HAM radio enthusiasts, trying to make the 8008 do something. Word had gotten around about what the Sac State machine could do but the cost of parts and degree of programming difficulty precluded most from pursuing this early technology. Only the fortuitous sharing of programming expertise and hardware resources shared between Sac State and Tektronix made that early working system possible. As mentioned above, also in 1973 Intel was developing the 8080 microprocessor and Bill's team received theirs in about December of that year. They quickly moved on to build another, much more comprehensive microcomputer setup to move more of their large medical application of the expensive host computer.
Curator Bruce Damer's note: Stan Mazor sent in the following first person impression on December 31, 2008:
Bill's career after the 8008 Bill also was involved in many other activities which eventually impacted the personal computer revolution. Later, Bill ended up being a facilitator who helped put the various people together to come up with the IMSAI 8080 to create a better design than the early Altair. One of Bill’s friends, Dallas Parcher a digital electronics technician for AT&T’s first digital switching network went on to repair the many early design and timing problems that plagued the early Altair and IMSAI 8080 personal computers. Bill meanwhile was earning his living as a professional host computer programmer and analyst. He designed and oversaw the development team who created a computer program that helped balance government budgets. When faced with the need to hand in a perfect departmental budget with no erasures or white out changes he designed and wrote a subvention system that became known as the Governor’s Budget Estimate System. That system used row numbers and column letters to identify unique cells. Every cell was composed of a value, character string, or formula. His system ran each set of values twice as it had to create an intermediate COBOL program to figure out the formulas. This system also allowed the user to setup automatic subtotaling based upon indentation of items in the first column. He later changed this system to use the Waterloo BASIC which allowed using IBM color terminals to make changes on the fly and instantly recalculate. This package was widely distributed by IBM under the names Budget Estimating system and MainCalc, a mainframe-based accounting application. It may well have been the first spread sheet that inspired VisiCalc (we don't know just yet) and much more. - Bruce Damer, Curator (with thanks to Bill Pentz for numerous corrections and additions) View Bill's interview by Allan Lundell and see photos of the visit below. We could also use your help to fill out this story more, see our Call for Help below. Bill's emails on his roles, history and the provenance of the donated artifacts (more writing from Bill to come):
Do you have any insight or background during this period either with SacState, Intel, Tektronix or in general, that could help shed some light on this story? Do you have a Tektronix 4023 or Memorex disk drive system you could donate for us to be able to put back together a semblance of the more complete pieces of the SacState 8008 system? Do you have any Popular Science magazines from 1973-74 you could donate or look through to see if you can find the little news bar item (Bill thinks it was in yellow) that talks about the SacState project with the 8008? Other Resources Related to the "SacState" Machine
Memorex 630 drive, possibly the one used by the SacState 8008 Machine (source Computer History Museum archives: sheet 22 of Memorex brochure)... or was it the...
Wikipedia's Pages on the Intel 8008 and Gary Kildall The MCS-8 User Manual (PDF) with 8008 Data Sheet at Classiccmp (local copy here) Herb Johnson's site The first floppy controller for CP/M and S-100? Dan Bricklin's VisiCalc page Our pages on the Homebrew Computer Club Our Altair 8800 computers Back to Digibarn Stories |
||||||||||||||
Please
send site comments to our Webmaster.
Please see our notices about the content of this site and its usage.
(cc) 1998-
Digibarn Computer Museum,
some rights reserved under this
Creative Commons license.
