On 2026-11-04, Rich <rich@example.invalid> wrote:
c186282 <c186282@nnada.net> wrote:
On 11/3/26 20:55, Woozy Song wrote:
Usenet has a lot of people who have been using it since the 1990s, so
presumably over 50 years old.-a I am a bit older, I remember 9-track
tapes, punch cards and acoustic coupler modems.
<waves hand> Me too!
Umm ... they were EIGHT-track tapes :-)
Idiot. Woozy's not referring to "8-track audio tapes".
Uh, I think he was pulling our legs - note the smiley.
He's referring to 9-track digital computer tapes used on
old IBM mainframes. They were 9-tracks:
https://en.wikipedia.org/wiki/9-track_tape
Eight data bits plus one parity bit per byte makes for a "9-track"
tape.
If you're really old you'll remember 7-track tapes (6 data bits
plus one parity bit, popular before 8-bit bytes came along).
I don't think I ever actually used one of them, but I did read
about them. I think some units even had a hardware converter
option that let you attach them to a machine that did 8-bit
bytes; for every three 8-bit bytes you sent them they'd write
four 6-bit bytes to the tape. Good for converting your old
tapes...
GIER ? Dansk.
Remembered the name but not the particulars. Had
to look it up.
48-bit words, mag-core 'main memory', DRUM
storage drive.
"The standard equipment comprises an 8-channel
paper tape reader reading 1000 characters per second,
an 8-channel paper tape punch punching 150 characters
per second, and a typewriter which is used for both
input and output"
https://datamuseum.dk/bits/30002477
Barely more than a glorified programmable calculator.
But, for the time ........
I'm sure it was of good use for science and industry.
GIER ? Dansk.
Remembered the name but not the particulars. Had
to look it up.
48-bit words, mag-core 'main memory', DRUM
storage drive.
"The standard equipment comprises an 8-channel
paper tape reader reading 1000 characters per second,
an 8-channel paper tape punch punching 150 characters
per second, and a typewriter which is used for both
input and output"
https://datamuseum.dk/bits/30002477
Barely more than a glorified programmable calculator.
But, for the time ........
I'm sure it was of good use for science and industry.
On 11/6/2026 21:51, c186282 wrote:
GIER ? Dansk.
Remembered the name but not the particulars. Had
to look it up.
48-bit words, mag-core 'main memory', DRUM
storage drive.
Actually 42 bits: 40 data bits, plus 2 flag bits.
In instruction words, the flag bits were named h (for half-word; i.e.
the word contained 2 instructions; the shorter instruction format did
not have an "index" part) and f (for floating; i.e. the instruction
would be addf/mulf/divf/subf rather than add/mul/div/sub).
The drum was organized in 40-word tracks, and was available in 320 track
or 960 track versions.
"The standard equipment comprises an 8-channel
paper tape reader reading 1000 characters per second,
an 8-channel paper tape punch punching 150 characters
per second, and a typewriter which is used for both
input and output"
https://datamuseum.dk/bits/30002477
The RC2000 paper tape reader actually could read up to 2000 characters
per second. It contained a 512 byte buffer memory. The more open space
in the buffer, the faster the paper tape would advance. It was very impressive to see it-a running at full speed!
The console included a small loudspeaker wired to the overflow bit in
the main accumulator register. This allowed the programmer/operator to
track the progress of the program as it worked it way through a set of loops.
Someone at RC wrote a music compiler that translated to finely tuned
small loops to generate the specific pitch for each note. It used the typewriter for percussion. The music rolls in our library spanned from
"F|+r Elise" to "Horst Wessel Lied".
Barely more than a glorified programmable calculator.
But, for the time ........
I'm sure it was of good use for science and industry.
Yes it was puny, but for 1961, it was very impressive. It had a full Algol-60 compiler (in 9 passes), written by Peter Naur (of Backus-Naur fame). 56 machines were built, a few are still running.
"You never forget your first!"
On 11/6/26 22:51, c186282 wrote:
GIER ? Dansk.
Remembered the name but not the particulars. Had
to look it up.
48-bit words, mag-core 'main memory', DRUM
storage drive.
"The standard equipment comprises an 8-channel
paper tape reader reading 1000 characters per second,
an 8-channel paper tape punch punching 150 characters
per second, and a typewriter which is used for both
input and output"
https://datamuseum.dk/bits/30002477
Barely more than a glorified programmable calculator.
But, for the time ........
I'm sure it was of good use for science and industry.
I associate it with Gier Algol. Brinch-Hansen? Worth?
On 11/7/2026 06:31, Peter Flass wrote:
On 11/6/26 22:51, c186282 wrote:
GIER ? Dansk.
Remembered the name but not the particulars. Had
to look it up.
48-bit words, mag-core 'main memory', DRUM
storage drive.
"The standard equipment comprises an 8-channel
paper tape reader reading 1000 characters per second,
an 8-channel paper tape punch punching 150 characters
per second, and a typewriter which is used for both
input and output"
https://datamuseum.dk/bits/30002477
Barely more than a glorified programmable calculator.
But, for the time ........
I'm sure it was of good use for science and industry.
I associate it with Gier Algol. Brinch-Hansen? Worth?
Naur - of Backus-Naur. - https://en.wikipedia.org/wiki/Peter_Naur
Peter Naur was an astronomer, who was first introduced to computers when
he was visiting Mount Palomar to use the 200-inch (5 meter) telescope.
On his return to Denmark, he joined the group that was becoming A/S Regnecentralen, which first built a tube machine (DASK) and later the
GIER system (transistors) and RC4000 (LSI ICs).
My first programming lessons were from Peter Naur.
CoPilot notes that "an interesting connection is that several European observatories in the 1960s used GIER computers from Regnecentralen for astronomical data reduction. The most famous example was the Hamburg Meridian Circle expedition in Perth, Australia, where a GIER processed stellar position measurements."
The RC2000 paper tape reader actually could read up to 2000 characters
per second. It contained a 512 byte buffer memory. The more open space
in the buffer, the faster the paper tape would advance. It was very impressive to see it running at full speed!
"You never forget your first!"
A lot of the old mini's - even mainframes - were
VERY SLOW by any modern standard. Slow enough to
be hilarious these days. CPU speed barely factored
in - the I/O speed with cards/tapes/etc used up
a huge amount of time.
c186282 <c186282@nnada.net> schrieb:
A lot of the old mini's - even mainframes - were
VERY SLOW by any modern standard. Slow enough to
be hilarious these days. CPU speed barely factored
in - the I/O speed with cards/tapes/etc used up
a huge amount of time.
Hence spooling.
https://www.youtube.com/watch?v=LL8hIRJs8Kg&list=PLsMFdolQPga5vhREBhhhXzpb4XXPd9Pc_&index=10
has a film of a 1968 meeting about an early introduction (not sure
if it actually was the first). A large room, and it was packed.
It was of large interest, obviously.
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were
-a-a-a VERY SLOW by any modern standard. Slow enough to
-a-a-a be hilarious these days. CPU speed barely factored
-a-a-a in - the I/O speed with cards/tapes/etc used up
-a-a-a a huge amount of time.
Hence spooling.
-a Spooling HELPS ... but, at some point, you've gotta
-a actually read/write to the shitty slow media. Also
-a spooling requires RAM ... which was in short supply
-a for a long time. I remember big cubes full of mag-core
-a memory ..... and it wasn't MUCH memory either.
On 11/8/2026 18:24, c186282 wrote:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were
-a-a-a VERY SLOW by any modern standard. Slow enough to
-a-a-a be hilarious these days. CPU speed barely factored
-a-a-a in - the I/O speed with cards/tapes/etc used up
-a-a-a a huge amount of time.
Hence spooling.
-a-a Spooling HELPS ... but, at some point, you've gotta
-a-a actually read/write to the shitty slow media. Also
-a-a spooling requires RAM ... which was in short supply
-a-a for a long time. I remember big cubes full of mag-core
-a-a memory ..... and it wasn't MUCH memory either.
Back in my University Computer Center days, a few of us attended a class
in London on debugging paradigms. On the first day, we introduced
ourselves and described our workplace's significant computer systems.
One of the attendeed was from IBM Grenoble, and as he described their
main system with a whopping 16MB of main memory, we all oohed and aahed
and wondered aloud how anyone could make productive use of that much
memory. But he explained that they easily used it all up with their
static analysis tools, their version control system, their auditing
programs making sure the programs adhered to their programming standards etc. We all shook our heads at how IBM clearly had too much money on
their hands if they could spend it on such frills!!
16 MEGABYTES ??? Wow !!! :-)
The old mag-core memory units were MUCH smaller.
On 11/8/26 21:34, Lars Poulsen wrote:
On 11/8/2026 18:24, c186282 wrote:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were
-a-a-a VERY SLOW by any modern standard. Slow enough to
-a-a-a be hilarious these days. CPU speed barely factored
-a-a-a in - the I/O speed with cards/tapes/etc used up
-a-a-a a huge amount of time.
Hence spooling.
-a-a Spooling HELPS ... but, at some point, you've gotta
-a-a actually read/write to the shitty slow media. Also
-a-a spooling requires RAM ... which was in short supply
-a-a for a long time. I remember big cubes full of mag-core
-a-a memory ..... and it wasn't MUCH memory either.
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory, we
all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version control
system, their auditing programs making sure the programs adhered to
their programming standards etc. We all shook our heads at how IBM
clearly had too much money on their hands if they could spend it on
such frills!!
-a 16 MEGABYTES ??? Wow !!!-a :-)
-a The old mag-core memory units were MUCH smaller.
-a Also EXPENSIVE as all hell ... hey, little ferrite
-a doughnuts hand-placed on wires .......
-a The coolest was 'rope memory' ... the doughnuts
-a hand-placed on long strands of wires. the old
-a Lunar Excursion Module computer used it. NASA
-a basically contracted Good Knitter grannies ...
The coolest was 'rope memory' ... the doughnuts
-a-a hand-placed on long strands of wires. the old
-a-a Lunar Excursion Module computer used it. NASA
-a-a basically contracted Good Knitter grannies ...
Wasn't the rope memory ROM for that thing?
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
A lot of the old mini's - even mainframes - were
VERY SLOW by any modern standard. Slow enough to
be hilarious these days. CPU speed barely factored
in - the I/O speed with cards/tapes/etc used up
a huge amount of time.
Hence spooling.
Spooling HELPS ... but, at some point, you've gotta
actually read/write to the shitty slow media. ...Also
spooling requires RAM ...
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory, we
all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version control system, their auditing programs making sure the programs adhered to
their programming standards etc. We all shook our heads at how IBM
clearly had too much money on their hands if they could spend it on
such frills!!
On 2026-11-09, c186282 <c186282@nnada.net> wrote:
16 MEGABYTES ??? Wow !!! :-)
The old mag-core memory units were MUCH smaller.
Yup. The 360/67 at the university had 3 (later 4)
core boxes. Each was the size of a large refrigerator
and held all of 256K.
16MB was the theoretical maximum amount of memory that you
could hang on a 360. I say "theoretical" because nobody
could afford anywhere near that. In the early '70s, I
remember reading an article in a trade rag that described
how IBM rocked the industry by slashing the price of a
megabyte of memory from $75,000 to a mere $15,000.
And now we pay less than a dollar a gigabyte. Amazing.
On 11/8/26 20:54, c186282 wrote:
On 11/8/26 21:34, Lars Poulsen wrote:
On 11/8/2026 18:24, c186282 wrote:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were
-a-a-a VERY SLOW by any modern standard. Slow enough to
-a-a-a be hilarious these days. CPU speed barely factored
-a-a-a in - the I/O speed with cards/tapes/etc used up
-a-a-a a huge amount of time.
Hence spooling.
-a-a Spooling HELPS ... but, at some point, you've gotta
-a-a actually read/write to the shitty slow media. Also
-a-a spooling requires RAM ... which was in short supply
-a-a for a long time. I remember big cubes full of mag-core
-a-a memory ..... and it wasn't MUCH memory either.
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory,
we all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version
control system, their auditing programs making sure the programs
adhered to their programming standards etc. We all shook our heads at
how IBM clearly had too much money on their hands if they could spend
it on such frills!!
-a-a 16 MEGABYTES ??? Wow !!!-a :-)
-a-a The old mag-core memory units were MUCH smaller.
-a-a Also EXPENSIVE as all hell ... hey, little ferrite
-a-a doughnuts hand-placed on wires .......
-a-a The coolest was 'rope memory' ... the doughnuts
-a-a hand-placed on long strands of wires. the old
-a-a Lunar Excursion Module computer used it. NASA
-a-a basically contracted Good Knitter grannies ...
Wasn't the rope memory ROM for that thing?
On 11/8/26 23:39, Charlie Gibbs wrote:
On 2026-11-09, c186282 <c186282@nnada.net> wrote:
16 MEGABYTES ??? Wow !!! :-)
The old mag-core memory units were MUCH smaller.
Yup. The 360/67 at the university had 3 (later 4)
core boxes. Each was the size of a large refrigerator
and held all of 256K.
Yep, remember 'em.
DID promote extreme code discipline though :-)
And from there I went to my first job, programming a Univac 9300
with 16K of memory. Discipline indeed!
According to c186282 <c186282@nnada.net>:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
A lot of the old mini's - even mainframes - were
VERY SLOW by any modern standard. Slow enough to
be hilarious these days. CPU speed barely factored
in - the I/O speed with cards/tapes/etc used up
a huge amount of time.
Hence spooling.
Spooling HELPS ... but, at some point, you've gotta
actually read/write to the shitty slow media. ...Also
spooling requires RAM ...
Depends how you do it. IBM 70xx mainframes read and wrote everything to and from
tape. Then there was a cheaper 1401 with a tape drive, card reader/punch and printer that did card->tape and tape->card and tape->print.
Later on a customer tied together a 7090 and a slower 7040 with a high-speed interconnect so the 7040 did all the I/O and job management, spooling card and
print to disk, and the 7090 just ran the jobs.
Lars Poulsen <lars@beagle-ears.com> writes:
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory, we
all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version control
system, their auditing programs making sure the programs adhered to
their programming standards etc. We all shook our heads at how IBM
clearly had too much money on their hands if they could spend it on
such frills!!
I was undergraduate when 360/67 arrived and hired originally full-time responsible for OS/360 (360/67, running as 360/65).
Then CSC came out to install (virtual machine) CP/67 (3rd after CSC
itself and MIT Lincoln Labs) and I mostly get to play with it during my weekend 48hr window. I then spend a few months rewriting pathlengths for running OS/360 in virtual machine. Bare machine test ran 322secs
... initially 856secs (CP67 CPU 534secs). After a few months I had CP67
CPU down from 534secs to 113secs. I then start rewriting the dispatcher/scheduler (dynamic adaptive resource manager/default fair
share scheduling policy), paging, adding ordered seek queuing (from
FIFO) and mutli-page transfer channel programs (from FIFO and optimized
for transfers/revolution, getting 2301 paging drum from 70-80 4k transfers/sec to channel transfer peak of 270). I was invited to spring
SHARE meeting in Houston for "official" CP67/CMS announcement. Six
months after univ initial CP/67 install, CSC was giving one week class
in LA. I arrive on Sunday afternoon and asked to teach the class, it
turns out that the people that were going to teach it had resigned the
Friday before to join one of the 60s CSC CP67 commercial online
spin-offs.
Then before I graduate, I was hired into small group in Boeing CFO
office to help with the formation of Boeing Computer Services
(consolidate all dataprocessing into independent business unit). When I graduate I join the IBM Cambridge Scientific Center that had 768kbyte
360/67 (104 pageable 4k pages, and I upgrade the page replacement
algorithm with my global "LRU" algorithm).
Then Grenoble Science Center got a 1mbyte 360/67 (155 pageable 4k pages) running CP67/CMS and modify page replacement to local "LRU" algorithm
from 60s academic literature. Both centers had similar CMS workload,
except I got 75-80 users with better response and throughput than
Grenoble running 35 users.
On 2026-11-10, c186282 <c186282@nnada.net> wrote:
On 11/8/26 23:39, Charlie Gibbs wrote:
On 2026-11-09, c186282 <c186282@nnada.net> wrote:
16 MEGABYTES ??? Wow !!! :-)
The old mag-core memory units were MUCH smaller.
Yup. The 360/67 at the university had 3 (later 4)
core boxes. Each was the size of a large refrigerator
and held all of 256K.
Yep, remember 'em.
DID promote extreme code discipline though :-)
And from there I went to my first job, programming a
Univac 9300 with 16K of memory. Discipline indeed!
On Tue, 10 Nov 2026 02:25:40 GMT, Charlie Gibbs wrote:
And from there I went to my first job, programming a Univac 9300
with 16K of memory. Discipline indeed!
The machine was considered more valuable than you were.
On 11/9/26 12:48, Lynn Wheeler wrote:
Lars Poulsen <lars@beagle-ears.com> writes:
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory, we
all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version control
system, their auditing programs making sure the programs adhered to
their programming standards etc. We all shook our heads at how IBM
clearly had too much money on their hands if they could spend it on
such frills!!
I was undergraduate when 360/67 arrived and hired originally full-time
responsible for OS/360 (360/67, running as 360/65).
Then CSC came out to install (virtual machine) CP/67 (3rd after CSC
itself and MIT Lincoln Labs) and I mostly get to play with it during my
weekend 48hr window. I then spend a few months rewriting pathlengths for
running OS/360 in virtual machine. Bare machine test ran 322secs
... initially 856secs (CP67 CPU 534secs). After a few months I had CP67
CPU down from 534secs to 113secs. I then start rewriting the
dispatcher/scheduler (dynamic adaptive resource manager/default fair
share scheduling policy), paging, adding ordered seek queuing (from
FIFO) and mutli-page transfer channel programs (from FIFO and optimized
for transfers/revolution, getting 2301 paging drum from 70-80 4k
transfers/sec to channel transfer peak of 270). I was invited to spring
SHARE meeting in Houston for "official" CP67/CMS announcement. Six
months after univ initial CP/67 install, CSC was giving one week class
in LA. I arrive on Sunday afternoon and asked to teach the class, it
turns out that the people that were going to teach it had resigned the
Friday before to join one of the 60s CSC CP67 commercial online
spin-offs.
Then before I graduate, I was hired into small group in Boeing CFO
office to help with the formation of Boeing Computer Services
(consolidate all dataprocessing into independent business unit). When I
graduate I join the IBM Cambridge Scientific Center that had 768kbyte
360/67 (104 pageable 4k pages, and I upgrade the page replacement
algorithm with my global "LRU" algorithm).
Then Grenoble Science Center got a 1mbyte 360/67 (155 pageable 4k pages)
running CP67/CMS and modify page replacement to local "LRU" algorithm
from 60s academic literature. Both centers had similar CMS workload,
except I got 75-80 users with better response and throughput than
Grenoble running 35 users.
-a The newbies won't be much interested in any of
-a this 'ancient shit' ... have NO clue as to where
-a it all CAME from, and thus NO inspirations for
-a the future. The AI will tell them what/how to
-a think from now on, until it makes them obsolete.
-a By reports, the AIs can already self-program/evolve.
-a So far this is being prevented because it's SCARY.
-a IS scary.
-a They ain't US ... more like 'aliens'.
-a I see the 'philosophers' are STILL trying
-a to offer 'reasons' why they're not True
-a Consciousness ........ wow ......
-a Anthropic offered 100+ pages of info on why
-a they ARE basically 'beings' in its IPO
-a documents. Hope, and WARNINGS.
On 11/9/26 21:25, Charlie Gibbs wrote:
On 2026-11-10, c186282 <c186282@nnada.net> wrote:
On 11/8/26 23:39, Charlie Gibbs wrote:
On 2026-11-09, c186282 <c186282@nnada.net> wrote:
-a-a-a-a 16 MEGABYTES ??? Wow !!!-a :-)
-a-a-a-a The old mag-core memory units were MUCH smaller.
Yup.-a The 360/67 at the university had 3 (later 4)
core boxes.-a Each was the size of a large refrigerator
and held all of 256K.
-a-a-a Yep, remember 'em.
-a-a-a DID promote extreme code discipline though :-)
And from there I went to my first job, programming a
Univac 9300 with 16K of memory.-a Discipline indeed!
-a 16K ... very tight for a mainframe !
-a But, that's what there WAS back then.
-a UniVac, Honeywell, DEC, Burroughs etc ... they
-a all died for various reasons. Only IBM survives.
-a IBM has been the butt of many jokes - but clearly
-a they've done SOMETHING right. Wide diversification
-a and knowing WHEN to dump tech seems to be the
-a magic formula. Everybody from NVidia on down is
-a paying IBM royalties for basic tech. IBM mainframes,
-a not the same as the 60s models, are still being
-a sold. Would LOVE one or two of those big black
-a boxes ... not sure what I'd DO with them, but ...
-a Oh, most popular OS for those mainframes, the
-a IBM spin of RHEL :-)
-a Oh, if you're perverse, Plan-9 has also been--
-a ported to the IBM mainframes-a :-)
-a NOT a bad distributed-processing OS actually,
-a but decidedly non-mainstream. If M$ lawyers
-a finally DO figure out how to kill Linux then
-a we need ALTERNATIVES.
On 11/9/26 09:29, Peter Flass wrote:
On 11/8/26 20:54, c186282 wrote:
On 11/8/26 21:34, Lars Poulsen wrote:
On 11/8/2026 18:24, c186282 wrote:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were VERY SLOW >>>>>>> -a-a-a by any modern standard. Slow enough to be hilarious these >>>>>>> -a-a-a days. CPU speed barely factored in - the I/O speed with
-a-a-a cards/tapes/etc used up a huge amount of time.
Hence spooling.
-a-a Spooling HELPS ... but, at some point, you've gotta actually
-a-a read/write to the shitty slow media. Also spooling requires RAM >>>>> -a-a ... which was in short supply for a long time. I remember big
-a-a cubes full of mag-core memory ..... and it wasn't MUCH memory
-a-a either.
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory,
we all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version
control system, their auditing programs making sure the programs
adhered to their programming standards etc. We all shook our heads at
how IBM clearly had too much money on their hands if they could spend
it on such frills!!
-a-a 16 MEGABYTES ??? Wow !!!-a :-)
-a-a The old mag-core memory units were MUCH smaller.
-a-a Also EXPENSIVE as all hell ... hey, little ferrite doughnuts
-a-a hand-placed on wires .......
-a-a The coolest was 'rope memory' ... the doughnuts hand-placed on
-a-a long strands of wires. the old Lunar Excursion Module computer
-a-a used it. NASA basically contracted Good Knitter grannies ...
Wasn't the rope memory ROM for that thing?
Yes, ROM, not RAM. Not sure WHAT the L.E.M. had for RAM. 2K maybe ?
It's sure to be documented somewhere.
Try : https://en.wikipedia.org/wiki/Apollo_Guidance_Computer
As said, there was already much better tech but the govt contracting
process kinda had them locked in to 1962.
Apparently they DID have grannies weaving the rope-ROM to spec.
UniVac, Honeywell, DEC, Burroughs etc ... they
all died for various reasons. Only IBM survives.
On 11/9/2026 19:49, c186282 wrote:
On 11/9/26 21:25, Charlie Gibbs wrote:
And from there I went to my first job, programming a
Univac 9300 with 16K of memory.-a Discipline indeed!
-a 16K ... very tight for a mainframe !
-a But, that's what there WAS back then.
The Univac 9300 was sort of a cheapo 360/30. Where I met it, in 1972, it
was acting as a mini, ronning online spooling/local RJE for our Univac
1106, a role that other 1100-series installations had done with a 1004
"card processor".
https://archive.computerhistory.org/resources/text/Remington_Rand/Univac.1004.1964.102646252.pdf
https://www.pinterest.com/pin/404409241532785626/
http://bitsavers.informatik.uni-stuttgart.de/pdf/univac/1100/datapro/70C-877-11_7009_UNIVAC_1106_1108.pdf
On Tue, 10 Nov 2026 02:25:40 GMT, Charlie Gibbs wrote:
And from there I went to my first job, programming a Univac 9300
with 16K of memory. Discipline indeed!
The machine was considered more valuable than you were.
On 2026-11-10, Lawrence DrCOOliveiro <ldo@nz.invalid> wrote:
On Tue, 10 Nov 2026 02:25:40 GMT, Charlie Gibbs wrote:
And from there I went to my first job, programming a Univac 9300 with
16K of memory. Discipline indeed!
The machine was considered more valuable than you were.
Yup. Time on machines of that class was worth about $150 an hour.
Depends how you do it. IBM 70xx mainframes read and wrote everything to and from
tape. Then there was a cheaper 1401 with a tape drive, card reader/punch and >> printer that did card->tape and tape->card and tape->print.
Later on a customer tied together a 7090 and a slower 7040 with a high-speed >> interconnect so the 7040 did all the I/O and job management, spooling card and
print to disk, and the 7090 just ran the jobs.
I'll have to try and find the I/O speed specs
for that setup.
My guess is that it was just *awful* ...
In theory drum memory COULD have been used
for spooling -
On 11/9/26 19:55, c186282 wrote:
On 11/9/26 12:48, Lynn Wheeler wrote:
Lars Poulsen <lars@beagle-ears.com> writes:
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as
he described their main system with a whopping 16MB of main memory, we >>>> all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily
used it all up with their static analysis tools, their version control >>>> system, their auditing programs making sure the programs adhered to
their programming standards etc. We all shook our heads at how IBM
clearly had too much money on their hands if they could spend it on
such frills!!
I was undergraduate when 360/67 arrived and hired originally full-time
responsible for OS/360 (360/67, running as 360/65).
Then CSC came out to install (virtual machine) CP/67 (3rd after CSC
itself and MIT Lincoln Labs) and I mostly get to play with it during my
weekend 48hr window. I then spend a few months rewriting pathlengths for >>> running OS/360 in virtual machine. Bare machine test ran 322secs
... initially 856secs (CP67 CPU 534secs). After a few months I had CP67
CPU down from 534secs to 113secs. I then start rewriting the
dispatcher/scheduler (dynamic adaptive resource manager/default fair
share scheduling policy), paging, adding ordered seek queuing (from
FIFO) and mutli-page transfer channel programs (from FIFO and optimized
for transfers/revolution, getting 2301 paging drum from 70-80 4k
transfers/sec to channel transfer peak of 270). I was invited to spring
SHARE meeting in Houston for "official" CP67/CMS announcement. Six
months after univ initial CP/67 install, CSC was giving one week class
in LA. I arrive on Sunday afternoon and asked to teach the class, it
turns out that the people that were going to teach it had resigned the
Friday before to join one of the 60s CSC CP67 commercial online
spin-offs.
Then before I graduate, I was hired into small group in Boeing CFO
office to help with the formation of Boeing Computer Services
(consolidate all dataprocessing into independent business unit). When I
graduate I join the IBM Cambridge Scientific Center that had 768kbyte
360/67 (104 pageable 4k pages, and I upgrade the page replacement
algorithm with my global "LRU" algorithm).
Then Grenoble Science Center got a 1mbyte 360/67 (155 pageable 4k pages) >>> running CP67/CMS and modify page replacement to local "LRU" algorithm
from 60s academic literature. Both centers had similar CMS workload,
except I got 75-80 users with better response and throughput than
Grenoble running 35 users.
-a-a The newbies won't be much interested in any of
-a-a this 'ancient shit' ... have NO clue as to where
-a-a it all CAME from, and thus NO inspirations for
-a-a the future. The AI will tell them what/how to
-a-a think from now on, until it makes them obsolete.
-a-a By reports, the AIs can already self-program/evolve.
-a-a So far this is being prevented because it's SCARY.
-a-a IS scary.
-a-a They ain't US ... more like 'aliens'.
-a-a I see the 'philosophers' are STILL trying
-a-a to offer 'reasons' why they're not True
-a-a Consciousness ........ wow ......
-a-a Anthropic offered 100+ pages of info on why
-a-a they ARE basically 'beings' in its IPO
-a-a documents. Hope, and WARNINGS.
They better watch out - the next thing they'll want to do is unionize,
and go on strike.
On Mon, 09 Nov 2026 21:19:59 -0500, c186282 wrote:
On 11/9/26 09:29, Peter Flass wrote:
On 11/8/26 20:54, c186282 wrote:
On 11/8/26 21:34, Lars Poulsen wrote:
On 11/8/2026 18:24, c186282 wrote:
On 11/8/26 03:08, Thomas Koenig wrote:
c186282 <c186282@nnada.net> schrieb:
-a-a-a A lot of the old mini's - even mainframes - were VERY SLOW >>>>>>>> -a-a-a by any modern standard. Slow enough to be hilarious these >>>>>>>> -a-a-a days. CPU speed barely factored in - the I/O speed with >>>>>>>> -a-a-a cards/tapes/etc used up a huge amount of time.
Hence spooling.
-a-a Spooling HELPS ... but, at some point, you've gotta actually >>>>>> -a-a read/write to the shitty slow media. Also spooling requires RAM >>>>>> -a-a ... which was in short supply for a long time. I remember big >>>>>> -a-a cubes full of mag-core memory ..... and it wasn't MUCH memory >>>>>> -a-a either.
Back in my University Computer Center days, a few of us attended a
class in London on debugging paradigms. On the first day, we
introduced ourselves and described our workplace's significant
computer systems. One of the attendeed was from IBM Grenoble, and as >>>>> he described their main system with a whopping 16MB of main memory,
we all oohed and aahed and wondered aloud how anyone could make
productive use of that much memory. But he explained that they easily >>>>> used it all up with their static analysis tools, their version
control system, their auditing programs making sure the programs
adhered to their programming standards etc. We all shook our heads at >>>>> how IBM clearly had too much money on their hands if they could spend >>>>> it on such frills!!
-a-a 16 MEGABYTES ??? Wow !!!-a :-)
-a-a The old mag-core memory units were MUCH smaller.
-a-a Also EXPENSIVE as all hell ... hey, little ferrite doughnuts
-a-a hand-placed on wires .......
-a-a The coolest was 'rope memory' ... the doughnuts hand-placed on
-a-a long strands of wires. the old Lunar Excursion Module computer
-a-a used it. NASA basically contracted Good Knitter grannies ...
Wasn't the rope memory ROM for that thing?
Yes, ROM, not RAM. Not sure WHAT the L.E.M. had for RAM. 2K maybe ?
It's sure to be documented somewhere.
Try : https://en.wikipedia.org/wiki/Apollo_Guidance_Computer
As said, there was already much better tech but the govt contracting
process kinda had them locked in to 1962.
Apparently they DID have grannies weaving the rope-ROM to spec.
Excellent lecture here, that I attended last month:
https://youtu.be/vVqGy5cd1Uw?si=LnqF80ExKKtiseAq
c186282 <c186282@nnada.net> writes:
UniVac, Honeywell, DEC, Burroughs etc ... they
all died for various reasons. Only IBM survives.
I'm sure that Unisys (nee Burroughs) would dispute that assertion.
Excellent lecture here, that I attended last month:
https://youtu.be/vVqGy5cd1Uw?si=LnqF80ExKKtiseAq
Immediately goes to a "not a robot" thing.
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 56:23:08 |
| Calls: | 1,265 |
| Calls today: | 1 |
| Files: | 1,362 |
| D/L today: |
3 files (2,684K bytes) |
| Messages: | 285,806 |