18 Nov
1998
18 Nov
'98
2:19 a.m.
Date: Tue, 17 Nov 1998 21:15:23 -0600
From: Jay Jaeger <cube(a)msn.fullfeed.com>
To: classiccmp(a)u.washington.edu
Subject: Re: PDP 11/55 / 11/45 Prints (Was PDP 11/70 console wiring
diagram)
Message-ID: <199811180322.VAA14831(a)fullfeed.msn.fullfeed.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
"Things are seldom what they seem; skim milk masquerades as cream!"
Well, it turns out that my 11/45 is something of a mixed bag, and probably
isn't simply an 11/45 at all. The aluminum plate on the console
indicates KB11A S/N > 01-05231 . The sticker (mostly overlaid by another)
on top of the backplane indicates system S/N 1525.
But the overlay sticker is probably what counts. It indicates KB11D Rev.
D, Backplane Rev R,. System S/N 293. If I am reading things right, I
have an 11/55 in an 11/45 case.
I had never bothered to check before. Sigh. Kind of a pity that it isn't
itself anymore. The real pity is that UW Chem E. used to have a real
11/55 (white front panel), that I didn't pursue because I was out of
room, several years ago.
Sounds to me like you may have an late model 11/45.
My memory may be hazy... and my limited docs are in the crawl space until I
find one of these bad boys for my house.
Most of the KB11A's became KB11D's via an ECO.
I believe this ECO was needed for Unix and some RSX versions.
I believe the 11/70 KB11B's all became KB11C's as well.
There's no difference between the late 11/45's and the rest of the
KB11 line. The KB11A and KB11D were 11/45, 11/50 and 11/55's
and the KB11B and KB11C (and KB11CM) were 11/70's and 11/74 stuff
reconfigured for DEC use as 11/70's internally when they sold AT&T all
their remaining 11/70's for Telco use about 1983 or so.
I thought the difference between the 11/45, 11/50 and 11/55 was the memory
configuration. The 11/55 was semiconductor fastbus memory and the 11/45
was usually an all-core configuration. (Realistically they're
interchangeable... I'd love to see someone make a 128kb bipolar memory
retrofit with a standard 512k cache SIMM for the 11s...)
Imagine ECC and high speed unibus memory... All of RSX or RSTS
living in one memory board.
The 11/50 was MOS memory IIRC.
I believe the MOS and core were about 250ms cycle times...and the
bipolar was half that. (this may be totally wrong... the memory's real
hazy here...)
The fast 32k of bipolar memory in the 11/55 was usually configured to exist
at either the OS memory area (so the OS would be kind of running
out of fast "cache-like" memory or where the main application
lived for number crunching apps. The 11/55, I think, correctly
configured held the DEC speed record on Fortran crunching (faster than
the 11/70 which had a real cache).
The rest of the 11/55 memory was either standard core or mos.
Bill
18 Nov
18 Nov
11:45 a.m.
I am pretty sure that it was indeed a field upgrade (ECO, if you like),
since the board location chart with the KB11-D designation is clearly
overlaying the original (a very nice professional job of it, too 8-) .
The KB11-D was, according the literature I have, strictly an 11/55
designation. Since the 11/45 and 11/55 apparently used the same board set
(memory architecture was the difference, as you suggest), it would make my
machine, essentially, an 11/55. Unix would not have required it, nor would
RSX.
However, my machine's slots for the fast bipolar memory are all empty, so
it would, in fact, essentially still be an 11/45 for all intents and purposes.
At 11:19 AM 11/18/98 -0500, you wrote:
> Date: Tue, 17 Nov 1998 21:15:23 -0600
> From: Jay Jaeger <cube(a)msn.fullfeed.com>
> To: classiccmp(a)u.washington.edu
> Subject: Re: PDP 11/55 / 11/45 Prints (Was PDP 11/70 console wiring
> diagram)
> Message-ID: <199811180322.VAA14831(a)fullfeed.msn.fullfeed.com>
> Mime-Version: 1.0
> Content-Type: text/plain; charset="us-ascii"
>
> "Things are seldom what they seem; skim milk masquerades as cream!"
>
> Well, it turns out that my 11/45 is something of a mixed bag, and probably
> isn't simply an 11/45 at all. The aluminum plate on the console
> indicates KB11A S/N > 01-05231 . The sticker (mostly overlaid by
another)
on top of the
backplane indicates system S/N 1525.
But the overlay sticker is probably what counts. It indicates KB11D Rev.
D, Backplane Rev R,. System S/N 293. If I am reading things right, I
have an 11/55 in an 11/45 case.
I had never bothered to check before. Sigh. Kind of a pity that it isn't
itself anymore. The real pity is that UW Chem E. used to have a real
11/55 (white front panel), that I didn't pursue because I was out of
room, several years ago.
Sounds to me like you may have an late model 11/45.
My memory may be hazy... and my limited docs are in the crawl space until I
find one of these bad boys for my house.
Most of the KB11A's became KB11D's via an ECO.
I believe this ECO was needed for Unix and some RSX versions.
I believe the 11/70 KB11B's all became KB11C's as well.
There's no difference between the late 11/45's and the rest of the
KB11 line. The KB11A and KB11D were 11/45, 11/50 and 11/55's
and the KB11B and KB11C (and KB11CM) were 11/70's and 11/74 stuff
reconfigured for DEC use as 11/70's internally when they sold AT&T all
their remaining 11/70's for Telco use about 1983 or so.
I thought the difference between the 11/45, 11/50 and 11/55 was the memory
configuration. The 11/55 was semiconductor fastbus memory and the 11/45
was usually an all-core configuration. (Realistically they're
interchangeable... I'd love to see someone make a 128kb bipolar memory
retrofit with a standard 512k cache SIMM for the 11s...)
Imagine ECC and high speed unibus memory... All of RSX or RSTS
living in one memory board.
The 11/50 was MOS memory IIRC.
I believe the MOS and core were about 250ms cycle times...and the
bipolar was half that. (this may be totally wrong... the memory's real hazy here...)
The fast 32k of bipolar memory in the 11/55 was usually configured to exist
at either the OS memory area (so the OS would be kind of running
out of fast "cache-like" memory or where the main application
lived for number crunching apps. The 11/55, I think, correctly
configured held the DEC speed record on Fortran crunching (faster than
the 11/70 which had a real cache).
The rest of the 11/55 memory was either standard core or mos.
Bill
9533
days inactive
9533
days old
1 comments
2 participants
participants (2)
-
cube＠msn.fullfeed.com -
pechter＠shell.monmouth.com