PDA

View Full Version : What mainframes were really significant?



MattCarp
December 29th, 2006, 08:10 PM
I really am interested in the golden years of mainframe computing - the 1960s and 1970s. At this time, you'll recall that the major manufacturers were called Snow White and the Seven Dwarves, referring to IBM and Burroughs, Univac, NCR, Control Data, Honeywell, GE, and RCA.

Getting information on these original, old mainframes is relatively difficult. What I'm interested in knowing is what were the merits of the various mainframe families from these early pioneering companies? What did these companies compete on? Architecture? Serivce? Support? Price?

In the end, I'd like to deepen my understanding of the unique and significant architectures.

For example, it appears that the two RCA families - the Spectra and the RCA Series - were simply IBM 360 and 370 compatibles. Not interesting.

I can't find much about NCR and Honeywell (prior to the acquisition of GE's computer business)

Wikipedia has some good info on the GE-635, which seems to indicate they're only slightly interesting from the standpoint of being an early SMP machine and having programmable I/O controllers. But I'm wondering if the GE machines really were significant.

Control Data seems to have some good information available about their 6600 series - the first supercomputer. I'd put them in the interesting/significant camp.

Univac is company that arguable started the industry, so I'd certainly call them intersesting & significant.

Burroughs seemed to have a good reputatation and a unique stack-based design. Their OS (the "Master Control Program") gets positive comments, so I'd guess that was interesting & significant.

So, from the list, it seems like CDC, Univac, and Burroughs may be the most interesting from the group?

chuckcmagee
December 30th, 2006, 01:57 AM
Burroughs had a EXCELLENT OPERATING SYSTEM! Their problem was they just could NOT get their hardware act together. I worked on B90 and B1900 systems in the early 80s. The "CE" (customer engineer) was there so often that it became a joke around the office. He was always there repairing floppy drives or fixing something else that broke. The MCP was an excellent OS. It had multi-tasking and priority paging, memory protection, tons of modern stuff. It would do some bad things if you were stupid, like let you keep running more and more programs before the other were completed. It would run out of "real" memory to keep track of all those programs and freeze up. I never really researched why they finally went under. I know "they didn't go under, they got purchased" Ya, right.

tradde
January 9th, 2007, 11:34 AM
Burroughs had a EXCELLENT OPERATING SYSTEM! Their problem was they just could NOT get their hardware act together. I worked on B90 and B1900 systems in the early 80s. The "CE" (customer engineer) was there so often that it became a joke around the office. He was always there repairing floppy drives or fixing something else that broke. The MCP was an excellent OS. It had multi-tasking and priority paging, memory protection, tons of modern stuff. It would do some bad things if you were stupid, like let you keep running more and more programs before the other were completed. It would run out of "real" memory to keep track of all those programs and freeze up. I never really researched why they finally went under. I know "they didn't go under, they got purchased" Ya, right.

Actually Burroughs "purchased" Sperry and they merged to become Unisys.
I worked for Sperry from 1981 to 1986. Then for Unisys from 1986 to 2000.
I worked with some of the later V series machines. Burroughs and Sperry
machines were quite different from each other as were there markets.
I am not sure why Mike Blumenthal thought it would be good to combine
the two. :)

chuckcmagee
January 9th, 2007, 06:28 PM
Hmm, ok, they sunk together then.

Al Kossow
January 10th, 2007, 03:19 PM
> Getting information on these original, old mainframes is relatively difficult.

have you looked at
http://bitsavers.org/pdf ?

Woodym1
January 15th, 2007, 10:35 AM
The UNIVAC SS 80 and SS 90 were general purpose stored program business computers. Installed from the mid fifties into the late sixties. The 80 or 90 designation referred to it's card handling capability. That was either 80 column IBM Hollerith rectangular hole cards or the 90 column round hole card used in the earlier existing Remington tab equipment. The SS 80 used a card punch built by Bull of France. It was a monstrosity! Cams and gears and wheels and springs and a real maintenance headache.

The Solid State 80/90 was itself was a fairly successful kludge. A five thousand "word" storage drum. The actual logic was executed by magnetic amplifiers. The CPU logic ran on a 707 KHZ rectified AC at 1 KW power. This power came from six 2CX250B tubes in PP parallel. This tube is a ceramic grounded plate, forced air-cooled power tetrode. Negative 1700 VDC applied to the cathode, about - 1300 to the screen grids. The output frequency (707KHZ) was excited by a track on the storage drum. This system also had transistors, but mainly in a support function to the MagAmps.

Intermediate card image storage was provided by capacitors for the SS90 card punch. A drum line printer provided 132 columns width at 600 LPM output. Thyratrons were used to fire the print hammers.

The CPU had four registers C, A, L and X. Programming was essentially in machine language. One then had to worry about "latency" by placing the instructions on the drum so that as it rotated, the next instruction came up upon completion of the current instruction. A multi-pass program called X6 would provide much of this. A "word" was 10 BiQuinary bytes.

The last one I installed was pushed into the concrete floored computer room with a pick-up truck. This was at Virginia Stage Lines in Charlottesville, VA about 1962. I became quite adept in keeping those systems in operation, got so I could just listen to that Bull punch feed a card and immediately know if it was properly timed. Even over a standard voice telephone! I was responsible for a SS80 with an added magnetic tape system at FCPC at Damneck, VA, Naval Training Center until I left UNIVAC on 1/1/1968. This system was running about 99.6% uptime

Maybe around a hundred of them were installed. It's sales future was probably blown out of the water by the new IBM 1401.