CDC 6000s at Michigan State University

When I arrived at Michigan State University in the fall of 1974, their 6500 was the only academic mainframe on campus. There was a IBM 370/168 (?) in the separate administrative computer center, but it was never used for research or instruction. And while there were a few minicomputers around--such as some PDP-8's in Chemistry and some PDP-11's in Engineering--they were restricted to use by a given project or lab. Therefore, the CDC 6500 had to shoulder the computing burden for the vast majority of the roughly 50,000 students and faculty on campus. Given its modest power by today's standards, it did an amazingly good job.

The 6500 was a less-powerful successor to the famous 6600, which was introduced in 1964. (1964 is the date everyone quotes, but I don't think any 6600s were delivered until 1966.) The 6600 is sometimes referred to as the first RISC machine, due to its limited instruction set and generally uncomplicated design. (One wag said that RISC stood for "Really Invented By Seymour Cray".)

From a programmer's perspective, the 6500 was nearly identical to the 6600, except that it had two symmetric CPUs. Under the covers, each CPU was a traditional unified processor, as opposed to the more sophisticated 6600 with its multiple functional units. Like today's RISC machines, the 6600 could execute multiple instructions simultaneously. But the 6500 lacked this capability, making it a slower machine than the single-CPU 6600. In a world where, over a quarter of a century later, symmetric multiprocessing machines have yet to achieve much dominance, it's eye-opening that the 6500 successfully used SMP so long ago.

Michigan State bought its 6500 in 1968 to add interactive service and to supplement and eventually replace its CDC 3600. The 3600, which had been acquired in 1963, was the world's most powerful commercially-available computer at the time (I believe), but it certainly did not hold that title when it was retired and sold in 1974. I regret that I did not see the 3600 when I toured campus months before starting school; by the time I showed up for classes, the computer was gone. Incidentally, the 3600 was a 48-bit machine compared with the 24-bit 3300 I used in high school, and the computers were not closely related.

Though the 6500 was off-the-shelf hardware (aside from a custom interface to our network), the software we used was largely home-grown. The operating system was called SCOPE/Hustler. It was based on Control Data's batch-oriented SCOPE 3.2 (System Control Of Program Execution) OS. MSU's primary innovation was to add interactive service in a way that was well-integrated into the vendor's batch OS. "Hustler" was taken from the Paul Newman movie of the same name, which was popular when MSU was beginning its design of this homebrew OS. Evidently the programmers found OS data structure terminology--with queues (sounds like "cues"--get it?), tables, and pools--too tempting to resist the cute name. They even invented a "pocket" data structure. But the operating system lacked "balls".

By the mid-70's Control Data, a member of the original BUNCH competitors to IBM (Burroughs, Univac, NCR, Control Data, and Honeywell) had a less-than-robust share of the computer market. It may be surprising, then, that it supported several operating systems on its 6000 series. These included:

NOS & NOS/BE came about when 1) CDC decided to officially support KRONOS instead of leaving it a special project, and 2) CDC decided to start charging for its software. Software charging came after they "won" their anti-trust battle with IBM. Since they forced IBM to start charging for their OS, they felt they had to do the same thing. Many companies had contracts for free upgrades of SCOPE; so they changed the name to NOS/BE and started charging for it.

NOS/BE and NOS weren't too bad, but they came too late for some sites, especially universities. Hence, in addition to SCOPE/Hustler, there were other homebrew operating systems. I believe that Purdue's and the University of Texas at Austin each had their own variants of CDC OS's.

SCOPE/Hustler was not a complete rewrite of SCOPE 3.2. Care was taken to retain a high degree of compatibility so that vendor applications would run with little modification. We got full source code for SCOPE 3.4 and NOS/BE as well as CDC applications, so we could modify the vendor's applications or crib code from the latest OS release to get the latest compiler to run on our system.

Back then, our idea of an "application" was a compiler or OS utility; there weren't too many commercially available traditional applications for CDC systems. However, we did have such favorites as SPSS (a statistical package) and IMSL (FORTRAN math subroutine library). As I recall, these packages typically required small modifications to run on our system.

Control Data came out with four lines of computers compatible with the original 6600: the 6000 Series, the Cyber 70 Series, the Cyber 170 Series, and the Cyber 180 Series. (I include the 7600 and its successors in with these four lines, even though the 7600 had slightly different I/O.) There wasn't a whole lot of difference amoungst the 6000, 70, and 170 lines. I guess the marketting folks just decided it was time for a new line a couple of times. On the other hand, for the most part, the Cyber 180 Series really was a new line. Most of the Cyber 180's were dramatically different CISC machines that had a 6000 compatibility mode. (There were a few models in the 180 Series that were just the same old architecture with a new name again.)

By the mid-1970's, the 6500 (which had roughly the power of an Intel 386+387, as far as I can tell) was having a tough time keeping up with the campus computing demands. In extreme cases, the batch input queue had 1000 jobs in it. Large TV monitors were installed which displayed the status of the system, including (as I recall) the position of jobs in the input and output (i.e., print) queues. Students would cluster underneath these monitors just like passengers at an airport, waiting for their jobs to complete.

So around 1975, Michigan State purchased a used CDC 6400, a single-CPU computer whose CPU was identical to either of the 6500 CPUs. Many person-years were expended in enhancing the operating system so it could support multiple CPUs sharing access to peripherals. All this for a number of MIPs which could probably be had today on the used market for about US$ 400. When the work was complete, the 6400 was used as a production machine by Computer Laboratory employees during the day, and was used for testing in the evenings. This represented a significant improvement in quality-of-life for systems programmers who were used to very limited time slots in the wee hours of the morning.

Eventually, in December 1979, MSU purchased a Cyber 170-750, a single-CPU machine with about 5-7 times the CPU horsepower of either 6500 CPU. The decision was a controversial one, as many in the campus community wanted us to get a more popular computer, a computer which would have more software available for it and which would provide more professionally relevant experience for our students. In other words, an IBM.

Nevertheless, the 750 was heavily used. However, it represented the last major victory for CDC afiocioandos at our site. Around late 1984, after much discussion, MSU purchased a low-end IBM machine (a 4381-1), and a bottom-of-the-line Cyber 180/810. SCOPE/Hustler was modified to run on the 810, and the 810 ran our home-grown OS until the late 1980's.

Back to CDC 6500 frameset