CDC 750 Hardware Configuration, 1980

Here is the configuration of Michigan State's Control Data 170/750 system as of 1980, shortly after its installation. The 6500 system it replaced was still in operation, but is not shown here. (Starting around this time, the 6500 was used in a production mode by Computer Lab employees during the business day, and by systems programmers for testing at other times.)

Model Num Description
CDC 6634 1 Extended Core Storage (ECS). 500,000 60-bit words, and yes, it was really magnetic core (like the 6500 memory but unlike the 750 memory). Nominal transfer rate 10M words/sec.
CDC 808-2 (aka 6612) 1 Enormous, ancient disk drive with hydraulically-controlled heads. 132M 6-bit characters of storage; average access 107.5 ms; transfer rate 1.28 M chars/sec. The surprisingly high transfer rate comes from 12-bit parallel transfers: 12 heads transferring data at once. I've often wondered why modern drives don't do this.

By 1980, the 808 was used was used only for job swapping (when the ECS was full, or the job was being swapped for a "long" time; e.g., waiting for a tape mount). This allowed us to minimize the disk head movement and take full advantage of the high transfer rate.. As I recall, to work around the slow seek time, only a small range of cylinders were used, further decreasing its very low capacity/footprint ratio.

CDC 7054
& 7154
4 Disk controllers.
PDU 7054 1 Purdue disk controller. CDC charged a lot of money for its disk controllers, spawning a few attempts to create clones. I think MSU also tried to create its own disk controller, but it never succeeded. These attempts were generally a waste of time; the only way they could have made sense economically was if a sizeable number could have been sold to other institutions.
7 Removeable pack disk drive. 110 M 6-bit characters per pack; average access of 38.2 ms; nominal transfer rate of 921,600 chars/sec. Not all disk controllers could handle this transfer rate, so sometimes 1:2 sector interleaving was used.

We almost never removed any of these packs, even during "system time" (testing). I thought (and still think) it was a waste having removeable drives, but removeable drives were much more common back then.

Frankly, I'm surprised these old 844s had this big a capacity. Later, we purchased a few CDC 885 drives, fixed "Winchester" disks of higher capacity.

6 Removeable pack disk drive. 220 M 6-bit characters per pack; average access of 38.2 ms; nominal transfer rate of 921,600 chars/sec. A double-density version of the 844-2.
CDC 607 7 7-track tape drive. 150 inches/sec at 200, 566, or 800 chars/inch. This is the same model as I had used on the CDC 3300.
CDC 669 4 9-track tape drive. 200 ips at 800 or 1600 chars/inch. Initially, these units were very unreliable. The problem was worked around by instituting a very aggressive cleaning policy. I think that Operations Manager John Kohmetcher eventually decreed that the drives should be cleaned before each use.
CDC 405 3 Card reader; 1200 cards/minute. I don't know what we paid for these, but we sure got our money's worth: they were used constantly. They jammed fairly often, but they probably performed as reliably as you could expect for the high speed.
CDC 415 1 Non-printing card punch. 250 cards/minute. This was not used much, even in the 1970's.
CDC 512 2 Chain-style line printer, attached to mainframe channel. 1200 lines/minute. Used constantly. Reasonably good quality output.
DPC 9646 1 Data Printer Corp. line printer, attached to front-end. 400 lines/minute. This more-or-less industry standard printer was much less expensive than the proprietary 512, and had decent output quality. I believe it used the Data Products parallel interface that predated the now-common Centronics interface.
DPC 6466 2 Data Printer Corp. line printer, faster, uppercase-only version of above.
1 16-port communications multiplexor, 2000-4800 bps. Used for remote batch stations, known as EI200 (Export/Import) stations. There were several of these, located in departments such as Engineering and Chemistry. The EI200 sites had a card reader, a printer, and a console. Initially, EI200 stations were made of CDC proprietary hardware, but we eventually built our own based on Interdata 7/16 minicomputers.

Before the Interdata front end, we used the 6671 to handle terminal I/O, which was done through ADS (American Data Systems) boxes. ADS boxes were character division multiplexers; we did the demuxing in a PP. This was a much cheaper solution than CDC's official solution: a 6674. The later ADS boxes could handle 300 bps! I don't believe the 6674 ever did.  It took a minor modification of the 6671 hardware so that it would not throw away SYNc codes (which were used to say "this line doesn't have any data").

1 Front-end computer, running 100% homegrown software and attached to the 750 via a home-brew channel adapter. Interdata was acquired by instrument maker Perkin-Elmer subsequent to our purchase of the initial 7/32. In 1980, it had 20 110/300/1200 bps dialup lines via VADIC 3467 modems; 55 110/300 bps lines via VADIC 307C modems, and 48 hardwired lines. The 7/32 also controlled the Data Printer Corp. printers, which used the Data Products-style parallel interface.

Back to CDC 6500 frameset