home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!decwrl!infopiz!mccall!info-pdp11-newsgate!list
- Newsgroups: vmsnet.pdp-11
- Subject: Re: UNIX for 18-bit pdp-11:s...
- Message-ID: <9209130822.AA15000@WLV.IIPO.GTEGSC.COM>
- From: <sms@wlv.iipo.gtegsc.com>
- Date: Sun, 13 Sep 92 01:22:22 -0700
- Organization: The Internet
- Return-Path: <pat+@transarc.com>
- CC: info-pdp11@transarc.com
- Lines: 100
-
- > From: Chris Petrilli <petrilli@gnu.ai.mit.edu>
- > On anything less than a high-end PDP (ie. 11/70, 11/84, etc..) I think
-
- Sounds like a "high-end" pdp-11 is anything that _excludes_
- split I/D and supervisor mode. Guess that leave the 11/44, 11/53,
- 11/73, 11/83 out. 11/44s are _dirt_ cheap these days and run Unix
- just fine. 11/84s are also "cheap" (but cheap != free). Heck,
- even the 11/45 (ca. 1975) had those capabilities. Only the "low
- end" models left them out.
-
- A 70 is a "labor of love" these days - it was the last of the
- "real computer" since it had a front panel mit lights and switches.
- Gads i miss having a 70 but the house wiring and my budget can't
- stand one. Super machine when it was working right, but heaven
- help you if anything ever went wrong with one.
-
- > RSX is a much better route. Even RSTS would be a better choice than
- > Unix.
-
- Depends on what a person plans to do. RSX and RSTS were hand
- crafted in assembly to do their jobs well, and they do it very
- well indeed. I used to work with RSX/IAS (in MACRO-11 of course)
- and i don't miss at all the UPPERCASEFILENAMES and 6+3 character
- file names, lack of IP/TCP networking, effort to OPEN$ a file
- (gads, setting up a FDB was a chore), etc.
-
- > I have played with various Unix implementations on the 22bit
- > machines, and while usable, I just don't think that it's going to cut
-
- How recently? Tried 2.11BSD at all? It might be a bit different
- than earlier experiences with Unix on a pdp-11. I know V7 was
- a real PITA (especially on a 11/23 - had enough room left over
- after the kernel to run 1 or 2 user processes, the system spent
- most of its time swapping).
-
- > it for most people. I guess working on a 50+MIPS RS/6000 spoils me,
- > but I still enjoy playing with older machines. However, the problems
-
- It's spoiled almost everyone except the really hardcore folks who
- know that "good computing" (like good cooking) takes time ;-)
-
- *start soapbox*
-
- Why bring up a superdooper workstation and then confess to a
- fondness for old machines? I'm tempted to wager that a goodly
- percentage of the 50mips is doing nothing more than driving the
- glitzy displays... Over the years i've noticed machines getting
- faster, the screens more colourful and glitzy, but not a whole
- lot more work getting done.
-
- One does not buy (or accept for free) a pdp-11 because of it's
- blinding speed (although the 11/93 introduced in 1990 is quite
- fast for an 11). Speed comparisons between almost anything
- against a 11/44 or 11/73 would show the 11 as a slower machine.
- So what? I kinda like taking a watercloset break once in a while ;-)
-
- * end soapbox *
-
- > come about in that V7 and early BSD (pre 4.2) are not well supported
- > any more by PD code, much less good compilers.
-
- I have to disagree somewhat. V7 and descendants were never PD
- in the first place (the DEATHSTAR you know). Unix for the pdp-11
- will never be PD (or easily affordable) given USL's current
- attitude, etc. Current PD code takes at least a "high end" pdp-11
- to even attempt a port (and forget anything that comes out of the
- GNU project - it typically takes more memory than you can put on a 11).
-
- Agreed, the earlier compilers (the DECUS C compiler is based on
- a ~V7'ish version) were rather poor, but that reflects the era
- rather than the machine/system. Again, noone's done much - they're
- all "spoiled" and can't be bothered with less than ~50 mips machines.
-
- > My experience with the C compiler that is on the PDP is pretty bad.
-
- How recent? Granted noone's been knocking down the doors demanding
- to do a better compiler lately, but the current C compiler isn't
- all bad -- it's rather simple minded for sure, the optimizer cleans
- up after it to a great extent -- enums, structure passing, etc are
- present, seems to have all the comforts of home to me. It's not
- ANSI though, thankfully (function prototypes are _ugly_).
-
- > Last I recall, not only does it have quite a few bugs but it also
- > generates pretty bad code.
-
- No bugs here. Everything from the kernel (which is 4.3BSD ported
- over with _minimal_ changes - 'long' vs. 'int' stuff, usw.) to
- 'rn', to C-Kermit5A(183) compiles just fine (but again you need a
- "high end" pdp-11 to do it.
-
- > I don't want to discourage anyone, just tell them what they are getting
- > themselves into.
-
- Can't discourage me - i'm a hardcore 11'er :-) What they'd be
- getting into is some serious $$$ though, pdp-11 peripherals are
- not in the "PC" price range - it takes a certain "obsession" level
- to have a pdp-11 these days.
-
- Steven M. Schultz
- sms@wlv.iipo.gtegsc.com
-