15 Jul
2015
15 Jul
'15
10:28 a.m.
From: Sean Caron
Many examples of blinkenlights eye candy throughout
computer history
It wasn't _just_ eye candy; it was a real help in problem debugging (when the
machine was stopped), and you could tell a lot about what the machine was
doing (when it was running) from the way the lights changed.
When the overall machine cost came down, they were too expensive to be worth
what they cost, though.
Speaking of lights for feedback, anyone remember the 'run bar' - or whatever
they called it, my memory fails me - on the display on the Lisp Machines?
Actually, it was a series, IIRC - one for the CPU, one for the disk, etc. The
machine didn't have LEDs, but it used short lines on the bit-map display
instead.
IIRC, the idea was copied from the Knight TV's on MIT-AI. (Which I believe
were the first-ever bit-mapped displays - anyone know of an earlier one?)
Noel
15 Jul
15 Jul
10:35 a.m.
On Jul 15, 2015, at 1:28 PM, Noel Chiappa <jnc at
mercury.lcs.mit.edu> wrote:
Absolutely. When DEC introduced the Remote Diagnostic Console for the 11/70, they started
deploying those internally. That makes sense. But in the RSTS/E development group, we
put our foot down and told them ?take it out and give back our blinkenlights? because for
OS development the ability to judge what the machine is doing, or spot strange behavior,
from the light patterns is quite important.
I can see several reasons for the disappearance of console lights. Cost, of course. The
fact that it wasn?t important for people running turnkey applications, i.e., much of the
customer base. And the switch from assembly language to higher level languages.
Then there was the very occasional early machine with no lights at all ? the CDC 6000
series is the one I can think of. But there you had the real time console status display,
which was even better ? updated just as fast but with a whole lot more information.
paul
From: Sean Caron
Many examples of blinkenlights eye candy
throughout computer history
It wasn't _just_ eye candy; it was a real help in problem debugging (when the
machine was stopped), and you could tell a lot about what the machine was
doing (when it was running) from the way the lights changed. 
11:03 a.m.
On 07/15/2015 10:35 AM, Paul Koning wrote:
Then there was the very occasional early machine with
no lights at
all ? the CDC 6000 series is the one I can think of. But there you
had the real time console status display, which was even better ?
updated just as fast but with a whole lot more information.
...and that counters Neil's assertion that lights were too expensive.
Cray didn't use lights, neither did CDC as well as other manufacturers.
What they used was (usually) a separate processor with diagnostic
capabilities.
Lights and switches paled in comparison to what an intelligent
diagnostic processor could do--you could see the state of I/O channels,
the P-counter and the job to which it was attached, modify the status of
just about anything in the system and--in some cases even tell the state
of the cooling system. Some allowed the operator to degrade system
memory, allowing normal work to proceed using part of memory while
performing diagnostic testing on the other part.
Lights are a quaint holdover from the 1950s and early 60s and really a
cheap alternative to getting system information. The console on a360/195
is an example of the technology carried to a ridiculous extreme. The
minicomputers of the 70s with their lights and switches being the last
holdout, mostly because of cost.
--Chuck
11:09 a.m.
-----Original Message-----
From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Chuck
Guzis
Sent: 15 July 2015 19:03
To: General at classiccmp.org; Discussion at classiccmp.org:On-Topic and Off-
Topic Posts
Subject: Re: Reproducing old machines with newer technology (Re: PDP-12 at
the RICM)
On 07/15/2015 10:35 AM, Paul Koning wrote:
I rec all being told by an IBM'er that Amdahl had a patent on having a service
processor, so IBM paid fees to Amdahl , but on investigation it looks like they had
several:-
http://patents.justia.com/assignee/amdahl-corporation?page=4
I also believe that Amdahl paid IBM for the use of the patents on Virtual memory. I wonder
who paid most....
.. and perhaps this explains why IBM mid-range boxes did not have service processors...
Dave
G4UGM
Then there was the very occasional early machine
with no lights at all
? the CDC 6000 series is the one I can think of. But there you had
the real time console status display, which was even better ? updated
just as fast but with a whole lot more information.
...and that counters Neil's assertion that lights were too expensive.
Cray didn't use lights, neither did CDC as well as other manufacturers.
What they used was (usually) a separate processor with diagnostic
capabilities.
Lights and switches paled in comparison to what an intelligent diagnostic
processor could do--you could see the state of I/O channels, the P-counter
and the job to which it was attached, modify the status of just about
anything in the system and--in some cases even tell the state of the cooling
system. Some allowed the operator to degrade system memory, allowing
normal work to proceed using part of memory while performing diagnostic
testing on the other part.
Lights are a quaint holdover from the 1950s and early 60s and really a cheap
alternative to getting system information. The console on a360/195 is an
example of the technology carried to a ridiculous extreme. The
minicomputers of the 70s with their lights and switches being the last
holdout, mostly because of cost.
--Chuck 
11:11 a.m.
I remember when U Wisconsin ECE got their PDP-11/20 and I saw DOS
FORTRAN get stuck for the very first time. I told the more senior
student who was responsible for getting things going, preparing
documentation, etc. that the machine was in a loop, and never coming
out. He laughed at me, claiming there was no way I could know that.
Bzzzzzt. Wrong. Tee Hee. That machine is now in my personal collection.
(PDP-11 DOS tended to scramble file system links and get stuck like that
- which inevitably required reloading the operating system disk - an
RC11 that still ran just fine when I last spun it up a year or so ago).
On 7/15/2015 12:35 PM, Paul Koning wrote:
...
It wasn't _just_ eye candy; it was a real
help in problem debugging (when the
machine was stopped), and you could tell a lot about what the machine was
doing (when it was running) from the way the lights changed.
Absolutely. When DEC introduced the Remote Diagnostic Console for the 11/70, they
started deploying those internally. That makes sense. But in the RSTS/E development
group, we put our foot down and told them ?take it out and give back our blinkenlights?
because for OS development the ability to judge what the machine is doing, or spot strange
behavior, from the light patterns is quite important.
paul
2:28 p.m.
On 7/15/15 10:28 AM, Noel Chiappa wrote:
Speaking of lights for feedback, anyone remember the
'run bar' - or whatever
they called it, my memory fails me - on the display on the Lisp Machines?
Actually, it was a series, IIRC - one for the CPU, one for the disk, etc. The
machine didn't have LEDs, but it used short lines on the bit-map display
instead.
"Short" here = ~1"
Yes, I see them all the time when I'm using my Symbolics machines. ;-)
TTFN - Guy
3448
days inactive
3448
days old
5 comments
6 participants
participants (6)
-
cclist＠sydex.com -
cube1＠charter.net -
dave.g4ugm＠gmail.com -
ggs＠shiresoft.com -
jnc＠mercury.lcs.mit.edu -
paulkoning＠comcast.net