home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: alt.folklore.computers
- Path: sparky!uunet!stanford.edu!leland.Stanford.EDU!alderson
- From: alderson@elaine46.Stanford.EDU (Rich Alderson)
- Subject: Re: PDP ftp archive
- In-Reply-To: lasner@watsun.cc.columbia.edu (Charles Lasner)
- Message-ID: <1992Nov23.185607.15965@leland.Stanford.EDU>
- Originator: alderson@leland.Stanford.EDU
- Sender: news@leland.Stanford.EDU (Mr News)
- Reply-To: alderson@elaine46.Stanford.EDU (Rich Alderson)
- Organization: Stanford University Academic Information Resources
- References: <13632@ecs.soton.ac.uk> <1992Nov20.142937.9123@news.uiowa.edu> <1992Nov21.003656.17491@leland.Stanford.EDU> <1992Nov22.114337.15485@news.columbia.edu>
- Date: Mon, 23 Nov 92 18:56:07 GMT
- Lines: 96
-
- In article <1992Nov22.114337.15485@news.columbia.edu>, lasner@watsun (Charles Lasner) writes:
- >In article <1992Nov21.003656.17491@leland.Stanford.EDU> alderson@elaine46.Stanford.EDU (Rich Alderson) writes:
- >
- >>> PDP-6 36 DEC's first really big computer.
- >>
- >> (only 23 were ever built, cause KO hated
- >> big MIS-type machines)
- >
- >I heard it was 20, not 23.
-
- The number 23 was mentioned in the K. Olsen bio _The Ultimate Entrepreneur_. I
- think I also heard it from a friend who managed 36-bit engineering at DEC.
-
- >
- >>> PDP-10 1967 36 A PDP-6 upgrade, great timesharing.
- >>
- >> Not exactly--more a revival.
- >
- >Upgrade is an unfortunate word. I think the reference is to the fact that
- >they are instruction-set compatible, and that time-sharing didn't really
- >happen on a -6.
-
- Note the number of attribution arrows. *I* was objecting to calling it an
- upgrade. The *PDP-8 FAQ* calls it an upgrade.
-
- >> Hmmph. Leave it to the 12-bit crowd to get it wrong.
- >
- >Not as far wrong as these guys posting about "PDP's" as if -11's are the
- >only kind!
-
- Amen!
-
- >>
- >> The DEC-20 is the outgrowth of a smart decision by DEC: License the
- >> TENEX operating system written for the Dec-10 by Bolt, Beranek, and
- >> Newman, which featured real demand-paged virtual memory, and sell it
- >> as an *alternative to*, not a replacement for, Tops-10, which was the
- >> descendant of the *batch-processing* operating system on the PDP-6 and
- >> early PDP-10s.
- >
- >But the -20 *is* a -10. What's the difference between a KL-10 and the
- >high-end -20, the color of the cabinet? Yes, you don't run TOPS-20 on
- >a KA-10, but a KL can be called anything you want. The point is that model
- >numbers can be muddled, and clearly marketroids are the source of the
- >muddling. Calling them all -10's would've been fine, but...
-
- There are microcode differences between a DEC-20 and a Dec-10--the JSYS
- instruction, for example, as opposed to UUOs. The question is not what the
- *hardware* is--yes, a KL is a PDP-10 and can be either a DEC-20 or a Dec-10--
- but rather what the *12-bit* crowd knows about the differences.
-
- And you can't run TOPS-20 on either a KA or a KI, only on a KL.
-
- >Also, what about the ill-fated Jupiter? Wasn't that to be "doubled" again
- >and be renamed DECsystem-40?
-
- This requires a bit of digression on the model-numbering scheme. I don't know
- all the numbers for the Decsystem-10, but a couple of important ones are the
- 1050 (which gave its name to the Tops-10 compatibility package on TOPS-20), and
- the 1090 and 1091 (which were the high-end systems, differing in that the 1090
- used core and the 1091 used MOS memory--or so *I* was told by an FE friend.)
-
- (Yes, I was friends with my FE. Wanna make something of it? :-)
-
- The original DEC-20...
-
- (BTW, I tend to use the capitalizations common to manuals from Digital. Thus,
- the older system was the Decsystem-10 or DECsystem-10 and ran Tops-10, while
- the later system was the DECSYSTEM-20 and ran TOPS-20. Back to our story...)
-
- The original DEC-20 was the KL-10 model 2040.
-
- The next model was the 2050, which used a cache memory to speed up the demand-
- paged virtual memory.
-
- The final model was the 2060, which came in two flavours, depending on whether
- the box implemented "extended addressing"--that is, the original 18-bit address
- space was expanded to 30 bits in the Model B, while the Model A did not have
- this capability. The Model A would run versions of Tops-20 up to version 4;
- later versions would only run on the Model B.
-
- I said "final model" in the previous paragraph, but there are two things to
- add. The 2020 was a small machine, which was sort of a work-group version of
- the architecture. It did not have extended addressing, and ahd slightly
- different I/O instructions from the other -20's.
-
- The 2065 was an upgrade of the 2060--both models--with a larger, faster cache.
-
- Now, given all this, the original model number for the Jupiter was 2090.
- However, I was given a Sales Training Seminar binder for the never-introduced
- Jupiter which is labeled DECsystem-4050.
- --
- Rich Alderson 'I wish life was not so short,' he thought. 'Languages take
- such a time, and so do all the things one wants to know about.'
- --J. R. R. Tolkien,
- alderson@leland.stanford.edu _The Lost Road_
-