Discussion:
Computer clock speed up EC
(too old to reply)
Lew Pitcher
2023-05-06 17:10:38 UTC
Permalink
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.

Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
--
Lew Pitcher
"In Skills We Trust"
greymaus
2023-05-06 17:31:35 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s

I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover :)

It was the 386, I think.

There was a folklore story of a company that wanted to upgrade their
system, so a man came with a briefcase came, opened the machine, turned
a screw and left.
--
***@mail.com
Where is out money gone, Dude?
Bob Eager
2023-05-06 20:16:55 UTC
Permalink
Post by greymaus
Long ago, I was told (or I read) about an early business computer (I'm
guessing circa 1960) that the manufacturer sold in two different clock
speeds. If you bought the "slower" system, you could pay a licence fee
and have a CSR come to /remove/ a component that /limited/ the CPU
speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s
I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover
It was the 386, I think.
In the 1960s?

I remember IBM allegedly doing this on a pre-360 system.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Scott Lurndal
2023-05-07 16:10:55 UTC
Permalink
Post by greymaus
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s
I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover :)
It was the 386, I think.
Certainly nopt in 'circa 1960'.
m***@smaus.org
2023-05-08 10:16:58 UTC
Permalink
Post by Scott Lurndal
Post by greymaus
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s
I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover :)
It was the 386, I think.
Certainly nopt in 'circa 1960'.
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
Ahem A Rivet's Shot
2023-05-08 11:30:46 UTC
Permalink
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Scott Lurndal
2023-05-08 16:24:33 UTC
Permalink
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Sn!pe
2023-05-08 16:26:31 UTC
Permalink
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
--
^Ï^. – Sn!pe – My pet rock Gordon just is.

http://youtu.be/_kqytf31a8E
Scott Lurndal
2023-05-08 17:02:25 UTC
Permalink
Post by Sn!pe
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
Wikipedia is just a few keystrokes away. The answer to your question is no.
Sn!pe
2023-05-08 17:45:55 UTC
Permalink
Scott Lurndal <***@slp53.sl.home> wrote:

[...]
Post by Scott Lurndal
Post by Sn!pe
Wasn't the 8088 a math co-processor?
Wikipedia is just a few keystrokes away. The answer to your question is no.
Thank you for answering, Scott. Please don't bother to answer
anything else I might ask, I'll just check Wikipedia (top tip, BTW).
--
^Ï^. – Sn!pe – My pet rock Gordon just is.

http://youtu.be/_kqytf31a8E
gareth evans
2023-05-08 17:09:17 UTC
Permalink
Post by Sn!pe
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
That was the 8087.

8088 was an 8086 but with an 8 bit bus.

80186 / 80188 were the versions with some I/O built in.
Kerr-Mudd, John
2023-05-08 17:35:11 UTC
Permalink
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
Post by Sn!pe
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
and a few more instructions.
--
Bah, and indeed Humbug.
Ahem A Rivet's Shot
2023-05-08 17:47:34 UTC
Permalink
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Kerr-Mudd, John
2023-05-08 18:08:15 UTC
Permalink
On Mon, 8 May 2023 18:47:34 +0100
Post by Kerr-Mudd, John
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
Let's not forget MircoSloth's Operating Sytem names (at least they
went up, unless they went sideways) and Mail program name(s).
--
Bah, and indeed Humbug.
Charlie Gibbs
2023-05-08 19:33:58 UTC
Permalink
Post by Kerr-Mudd, John
On Mon, 8 May 2023 18:47:34 +0100
Post by Kerr-Mudd, John
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
Let's not forget MircoSloth's Operating Sytem names (at least they
went up, unless they went sideways)
And then stopped at 10 forever - or until 11 came long.
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Peter Flass
2023-05-09 00:30:55 UTC
Permalink
Post by Kerr-Mudd, John
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
--
Pete
Charlie Gibbs
2023-05-09 03:56:27 UTC
Permalink
Post by Peter Flass
Post by Kerr-Mudd, John
On Mon, 8 May 2023 18:09:17 +0100
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
I am Pentium of Borg.
Division is futile.
You will be approximated.
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Bob Eager
2023-05-09 07:43:36 UTC
Permalink
Post by Peter Flass
Post by Ahem A Rivet's Shot
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
No, that was the 80387.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Ahem A Rivet's Shot
2023-05-09 09:28:29 UTC
Permalink
On 9 May 2023 07:43:36 GMT
Post by Bob Eager
Post by Peter Flass
Post by Ahem A Rivet's Shot
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
No, that was the 80387.
See the maths was so bad they got the wrong chip.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Bob Eager
2023-05-09 07:45:41 UTC
Permalink
Post by Peter Flass
Post by Ahem A Rivet's Shot
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
No, the 80387. And not the early ones. The original 80386 (later named the
80386DX) had full maths support built in.

Then they did the 'economy' version, the 80386SX. Limited bus width, no
maths coprocessor. And that's where the 80387 came in.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Kerr-Mudd, John
2023-05-09 09:48:49 UTC
Permalink
On 9 May 2023 07:45:41 GMT
Post by Bob Eager
Post by Peter Flass
Post by Ahem A Rivet's Shot
Post by gareth evans
That was the 8087.
8088 was an 8086 but with an 8 bit bus.
80186 / 80188 were the versions with some I/O built in.
One thing Intel were always hopelessly bad at was numbering
conventions.
They did the math on an early 386.
No, the 80387. And not the early ones. The original 80386 (later named the
80386DX) had full maths support built in.
Then they did the 'economy' version, the 80386SX. Limited bus width, no
maths coprocessor. And that's where the 80387 came in.
I recall that being the 486sx - having the maths part on board but
disabled. (Same bus width though).

Tell you what let's read it on wikipedia.

i386:
https://en.wikipedia.org/wiki/80386SX
says both had co-processor addon possibilities - 80387/80387sx.
https://en.wikipedia.org/wiki/X87#80387

i486:
https://en.wikipedia.org/wiki/X87#80487
--
Bah, and indeed Humbug.
Ahem A Rivet's Shot
2023-05-09 12:03:49 UTC
Permalink
On 9 May 2023 07:45:41 GMT
Post by Bob Eager
No, the 80387. And not the early ones. The original 80386 (later named
the 80386DX) had full maths support built in.
Then they did the 'economy' version, the 80386SX. Limited bus width, no
maths coprocessor. And that's where the 80387 came in.
Er nope.

80386 - 32 bit no hardware maths support
80387 - Maths coprocessor for 80386
80386sx - 80386 with 16 bit external bus
80387sx - Maths coprocessor for 80386sx
80486dx - 32 bit with built in maths coprocessor - got up to 50MHz
80486sx - 32 bit without maths coprocessor
80486dx2 - 80486 with external bus running at half the internal clock speed

Then they decided that numbers were bad and produced the Pentium.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Scott Lurndal
2023-05-09 13:53:23 UTC
Permalink
Post by Kerr-Mudd, John
On 9 May 2023 07:45:41 GMT
Post by Bob Eager
No, the 80387. And not the early ones. The original 80386 (later named
the 80386DX) had full maths support built in.
Then they did the 'economy' version, the 80386SX. Limited bus width, no
maths coprocessor. And that's where the 80387 came in.
Er nope.
80386 - 32 bit no hardware maths support
80387 - Maths coprocessor for 80386
80386sx - 80386 with 16 bit external bus
80387sx - Maths coprocessor for 80386sx
80486dx - 32 bit with built in maths coprocessor - got up to 50MHz
80486sx - 32 bit without maths coprocessor
80486dx2 - 80486 with external bus running at half the internal clock speed
Then they decided that numbers were bad and produced the Pentium.
Internally, the codename was P5, which morphed into pentium branding.
The P6 morphed into Pentium Pro branding
The P7 was cancelled and replaced with Merced which became Itanium.

I used to have a yellow book for the original P7 when I was at Unisys.
Niklas Karlsson
2023-05-09 16:17:39 UTC
Permalink
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?

Niklas
--
"Vir! You are not biologically equipped to handle fast food."
"I know, I know, but it .. it tastes so great going down. Coming up again it's
not so terrific, but--"
-- Londo and Vir in Babylon 5:"Meditations on the Abyss"
richardthiebaud
2023-05-09 16:43:14 UTC
Permalink
Post by Niklas Karlsson
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?
Niklas
IIRC, there was a court ruling that you can't trademark a number.
Niklas Karlsson
2023-05-09 17:09:12 UTC
Permalink
Post by richardthiebaud
Post by Niklas Karlsson
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?
Niklas
IIRC, there was a court ruling that you can't trademark a number.
Ah, that rings a faint bell, so that's probably what I was thinking of.
Thanks!

Niklas
--
I am strongly leaning towards the theory that Microsoft is the
spearhead of an alien invasion, designed specifically to cripple Earth's
development of technology.
-- Mark 'Kamikaze' Hughes in asr
Ahem A Rivet's Shot
2023-05-09 17:21:06 UTC
Permalink
On Tue, 9 May 2023 12:43:14 -0400
Post by richardthiebaud
Post by Niklas Karlsson
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the
Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?
Niklas
IIRC, there was a court ruling that you can't trademark a number.
Hmm that would have irritated Peugeot who reportedly trademarked all
three digit numbers with a 0 in the middle. IIRC it was more subtle than
that.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Vir Campestris
2023-05-11 16:46:24 UTC
Permalink
Post by Ahem A Rivet's Shot
Hmm that would have irritated Peugeot who reportedly trademarked all
three digit numbers with a 0 in the middle. IIRC it was more subtle than
that.
And that would have irritated Boeing. Heck, I just booked a flight on a
787; that number derives from the 707 way back when.

Andy
Ahem A Rivet's Shot
2023-05-11 19:58:31 UTC
Permalink
On Thu, 11 May 2023 17:46:24 +0100
Post by Vir Campestris
Post by Ahem A Rivet's Shot
Hmm that would have irritated Peugeot who reportedly
trademarked all three digit numbers with a 0 in the middle. IIRC it was
more subtle than that.
And that would have irritated Boeing. Heck, I just booked a flight on a
787; that number derives from the 707 way back when.
I think Peugeot's trademark only applied to cars.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Johnny Billquist
2023-05-12 10:19:51 UTC
Permalink
Post by Ahem A Rivet's Shot
On Thu, 11 May 2023 17:46:24 +0100
Post by Vir Campestris
Post by Ahem A Rivet's Shot
Hmm that would have irritated Peugeot who reportedly
trademarked all three digit numbers with a 0 in the middle. IIRC it was
more subtle than that.
And that would have irritated Boeing. Heck, I just booked a flight on a
787; that number derives from the 707 way back when.
I think Peugeot's trademark only applied to cars.
True. Which annoyed Porsche, who had to rename their Porsche 901 to
Porsche 911.

Johnny
greymaus
2023-05-12 14:28:42 UTC
Permalink
Post by Johnny Billquist
Post by Ahem A Rivet's Shot
On Thu, 11 May 2023 17:46:24 +0100
Post by Vir Campestris
Post by Ahem A Rivet's Shot
Hmm that would have irritated Peugeot who reportedly
trademarked all three digit numbers with a 0 in the middle. IIRC it was
more subtle than that.
And that would have irritated Boeing. Heck, I just booked a flight on a
787; that number derives from the 707 way back when.
I think Peugeot's trademark only applied to cars.
True. Which annoyed Porsche, who had to rename their Porsche 901 to
Porsche 911.
Johnny
which might not have helped in the us market. What year was that?
--
***@mail.com
Where is our money gone, dude?
Ahem A Rivet's Shot
2023-05-12 17:06:37 UTC
Permalink
On 12 May 2023 14:28:42 GMT
Post by greymaus
which might not have helped in the us market. What year was that?
The Porche 911 has been around since the 1960s.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Johnny Billquist
2023-05-15 10:25:39 UTC
Permalink
Post by Ahem A Rivet's Shot
On 12 May 2023 14:28:42 GMT
Post by greymaus
which might not have helped in the us market. What year was that?
The Porche 911 has been around since the 1960s.
Yup. And if you look at the spare parts catalogs for the old models,
you'll see a lot of part numbers starting with 901.xxx.xxx
So internally, the 901 number stuck, but externally, they had to rename it.

(The current 911 is actually 997 or something.)

Johnny
Kurt Weiske
2023-05-12 14:40:00 UTC
Permalink
To: richardthiebaud
-=> richardthiebaud wrote to alt.folklore.computers <=-

ri> IIRC, there was a court ruling that you can't trademark a number.

Different field, but the Porsche 911 was originally called the 901.
Peugot took them to court claiming they had some right to 3 digit car
names with a 0 in the middle digit, or something like that. Porsche
changed it to 911.



... RAW DATA FOR RAW NERVES
--- MultiMail/Win v0.52
--- Synchronet 3.20a-Win32 NewsLink 1.114
* realitycheckBBS - Aptos, CA - telnet://realitycheckbbs.org
Rich Alderson
2023-05-09 22:37:38 UTC
Permalink
Post by Niklas Karlsson
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?
Actually, the digit string "586" was used by Intel to name their Ethernet
chips. No kerfuffle involved, just bad decisions earlier in their own history.
--
Rich Alderson ***@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen
Peter Flass
2023-05-10 22:53:16 UTC
Permalink
Post by Rich Alderson
Post by Niklas Karlsson
Post by Ahem A Rivet's Shot
Then they decided that numbers were bad and produced the Pentium.
Wasn't there some kerfuffle about "586" or the like already being
copyrighted by someone else?
Actually, the digit string "586" was used by Intel to name their Ethernet
chips. No kerfuffle involved, just bad decisions earlier in their own history.
IBM now seems go be re-using machine numbers. I search for some old piece
of gear and get a lot of newer stuff I could care less about.
--
Pete
Charlie Gibbs
2023-05-09 18:00:12 UTC
Permalink
Post by Kerr-Mudd, John
On 9 May 2023 07:45:41 GMT
Post by Bob Eager
No, the 80387. And not the early ones. The original 80386 (later named
the 80386DX) had full maths support built in.
Then they did the 'economy' version, the 80386SX. Limited bus width, no
maths coprocessor. And that's where the 80387 came in.
Er nope.
80386 - 32 bit no hardware maths support
80387 - Maths coprocessor for 80386
80386sx - 80386 with 16 bit external bus
80387sx - Maths coprocessor for 80386sx
80486dx - 32 bit with built in maths coprocessor - got up to 50MHz
80486sx - 32 bit without maths coprocessor
80486dx2 - 80486 with external bus running at half the internal clock speed
Then they decided that numbers were bad and produced the Pentium.
8086 One little
80286 Two little
80386 Three little-endians
80387 Four little
80386sx Five little
80387sx Six little-endians
80486dx Seven little
80486sx Eight little
80486dx2 Nine little-endians
Pentium DIVIDE ERROR
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Peter Flass
2023-05-09 00:30:53 UTC
Permalink
Post by Sn!pe
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
I think the 8088 was used in the Displaywriter. I believe it was faster
than the 8086 (at least the number would imply that), but I don’t know
what the other differences were.
--
Pete
Charlie Gibbs
2023-05-09 03:56:26 UTC
Permalink
Post by Peter Flass
Post by Sn!pe
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
On 8 May 2023 10:16:58 GMT
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
Wasn't the 8088 a math co-processor?
I think the 8088 was used in the Displaywriter. I believe it was faster
than the 8086 (at least the number would imply that), but I don’t know
what the other differences were.
If you swap 8088 and 8086 in your last paragraph you'll be exactly right.
What the numbers imply is not in fact correct.
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Bob Eager
2023-05-08 18:12:41 UTC
Permalink
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
And the lesser known 81016/80188, and 80286/80288.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Charlie Gibbs
2023-05-08 19:33:57 UTC
Permalink
Post by Bob Eager
Post by Scott Lurndal
Post by Ahem A Rivet's Shot
Post by m***@smaus.org
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
There was the 386 with a 32 bit bus and the 386sx with a 16 bit bus.
And, of course, the 8086 vs 8088.
And the lesser known 81016/80188, and 80286/80288.
Don't forget the Motorola 68000 and 68008
(16- and 8-bit buses, respectively).
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Bob Eager
2023-05-08 12:38:27 UTC
Permalink
Post by m***@smaus.org
Post by Scott Lurndal
Post by greymaus
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two different
clock speeds. If you bought the "slower" system, you could pay a
licence fee and have a CSR come to /remove/ a component that
/limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s
I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover :)
It was the 386, I think.
Certainly nopt in 'circa 1960'.
Not in 1960, I agree, but there was something about thhe 386 comin in 2 versions
I think he was confusing it with the 360!
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
D.J.
2023-05-07 16:47:34 UTC
Permalink
Post by greymaus
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
s
I have a distant memory of all that. It, as I recall, happened several
times, as even now you can pay more than the basic price for a shiny
cover :)
It was the 386, I think.
There was a folklore story of a company that wanted to upgrade their
system, so a man came with a briefcase came, opened the machine, turned
a screw and left.
The university I was at didn't have any 386 computers until about
1988. ran AutoCAD on them.
--
Jim
Dennis Boone
2023-05-06 18:09:57 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Since at least the 80s, many of IBMs machines have had more performance
built in than was necessarily available. Microcode replacements or
perhaps "magic" codes (crypto or whatever) have turned one model into a
higher performing one. In the 4381 family, when we did this to one of
ours, the IBM FE was required to kick us all out of the room to swap the
floppies containing the microcode.

These days, there's a complex licensing system that allows IBM to charge
by actual usage, or for customers to pay for processing power on demand.
If you need more oomph for a short term project, pay IBM more for those
hours, days or weeks. The machines phone home.

De
Peter Flass
2023-05-06 18:26:57 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
It was fairly common. I think the IBM 1130 worked this way. Smart customers
might remove the jumper, and replace it before scheduled maintenance.
--
Pete
Lew Pitcher
2023-05-06 18:50:31 UTC
Permalink
Post by Peter Flass
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
It was fairly common. I think the IBM 1130 worked this way. Smart customers
might remove the jumper, and replace it before scheduled maintenance.
Thanks, that might be the lead I'm looking for: the story I remember talked about
the CSR removing a jumper or wire to enable the clock speedup.

I'll have to do some serious googling to track down the rumour, but your comment
gives me a place to start.


Thanks again
--
Lew Pitcher
"In Skills We Trust"
Ahem A Rivet's Shot
2023-05-06 18:16:38 UTC
Permalink
On Sat, 6 May 2023 17:10:38 -0000 (UTC)
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
I recall the tale - when I heard it the component was a link that
set the clock rate.
Post by Lew Pitcher
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
It certainly does not seem unlikely, I recall seeing hard discs
where a link change could double the capacity (10Mb to 20Mb IIRC) - it was a
hack because the smaller ones were no longer available but service contracts
only allowed replacement not upgrade. The service engineer was usually
careful to mention that you should not change the link because then you'd
have to reformat the drive and reinstall the OS.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Lew Pitcher
2023-05-06 18:51:35 UTC
Permalink
Post by Ahem A Rivet's Shot
On Sat, 6 May 2023 17:10:38 -0000 (UTC)
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
I recall the tale - when I heard it the component was a link that
set the clock rate.
That agrees with my recollection of the rumour as well.

[snip]


Thanks for the confirmation :-)
--
Lew Pitcher
"In Skills We Trust"
Lew Pitcher
2023-05-06 18:58:59 UTC
Permalink
Post by Lew Pitcher
Post by Ahem A Rivet's Shot
On Sat, 6 May 2023 17:10:38 -0000 (UTC)
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
I recall the tale - when I heard it the component was a link that
set the clock rate.
That agrees with my recollection of the rumour as well.
I've found one reference to the Univac 1106, which (according to
the reference) could be "upgraded" to an 1108.
"This upgrade consisted in replacing a divide-by-two flip-flop
in the system clock circuit by a jumper."

This might be the rumour I was looking for, but the source is
not definitive. I'll keep looking
Post by Lew Pitcher
[snip]
Thanks for the confirmation :-)
--
Lew Pitcher
"In Skills We Trust"
Peter Flass
2023-05-06 21:15:35 UTC
Permalink
Post by Ahem A Rivet's Shot
On Sat, 6 May 2023 17:10:38 -0000 (UTC)
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
I recall the tale - when I heard it the component was a link that
set the clock rate.
Post by Lew Pitcher
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
It certainly does not seem unlikely, I recall seeing hard discs
where a link change could double the capacity (10Mb to 20Mb IIRC) - it was a
hack because the smaller ones were no longer available but service contracts
only allowed replacement not upgrade. The service engineer was usually
careful to mention that you should not change the link because then you'd
have to reformat the drive and reinstall the OS.
360/20 had, I believe, a physical stop that prevented seeking to half the
disk.
--
Pete
Sn!pe
2023-05-06 19:03:33 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
IME it was standard practise in the early '70s; at least, it was on the
kit I worked with. When the customer buys a computer he buys a black
box with certain capabilities. From the manufacturer's point of view
it makes sense to build the best box they reasonably can and throttle
it. When the customer needs more power, charge him more and just take
out the throttling link. Not needing to phsically change the box makes
all sorts of good sense for everybody.
--
^Ï^. – Sn!pe – My pet rock Gordon just is.


Lew Pitcher
2023-05-06 19:51:57 UTC
Permalink
Post by Sn!pe
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
IME it was standard practise in the early '70s; at least, it was on the
kit I worked with. When the customer buys a computer he buys a black
box with certain capabilities. From the manufacturer's point of view
it makes sense to build the best box they reasonably can and throttle
it. When the customer needs more power, charge him more and just take
out the throttling link. Not needing to phsically change the box makes
all sorts of good sense for everybody.
Apparently, it is becoming standard practice again. But, this time
in the field of consumer goods.

I asked about the rumour because I just finished reading an article
(https://www.techdirt.com/2023/05/05/mercedes-locks-better-ev-engine-performance-behind-annoying-subscription-paywalls)
about Mercedes planning to put "better engine performance" of their
EVs behind a subscription paywall.

You buy the car. If you want better engine performance, you pay
a subscription, and they /disable/ the "poorer engine performance"
component. In other words, the car that you purchased /already/ has
"better engine performance", and you pay a monthly ransom to get it
enabled.

The analogy struck me as apropos.
--
Lew Pitcher
"In Skills We Trust"
Sn!pe
2023-05-06 21:07:39 UTC
Permalink
Post by Lew Pitcher
Post by Sn!pe
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
IME it was standard practise in the early '70s; at least, it was on the
kit I worked with. When the customer buys a computer he buys a black
box with certain capabilities. From the manufacturer's point of view
it makes sense to build the best box they reasonably can and throttle
it. When the customer needs more power, charge him more and just take
out the throttling link. Not needing to phsically change the box makes
all sorts of good sense for everybody.
Apparently, it is becoming standard practice again. But, this time
in the field of consumer goods.
I asked about the rumour because I just finished reading an article
(https://www.techdirt.com/2023/05/05/mercedes-locks-better-ev-engine-performance-behind-annoying-subscription-paywalls)
about Mercedes planning to put "better engine performance" of their
EVs behind a subscription paywall.
You buy the car. If you want better engine performance, you pay
a subscription, and they /disable/ the "poorer engine performance"
component. In other words, the car that you purchased /already/ has
"better engine performance", and you pay a monthly ransom to get it
enabled.
The analogy struck me as apropos.
Indeed. It's the same with Teslas (for one example).
--
^Ï^. – Sn!pe – My pet rock Gordon just is.

http://youtu.be/_kqytf31a8E
Charlie Gibbs
2023-05-07 05:02:34 UTC
Permalink
Post by Lew Pitcher
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
The Univac 9300, introduced in 1965, was their answer to the IBM 360/20.
Its 16-bit memory addressing was almost but not exactly unlike that of
the 360/20, but it executed instructions at the speed of a 360/30
(for the limited subset of instructions that it supported).

To cater to the low end of the market, they also released the 9200,
a de-rated version of the 9300. Some changes were physical: the
printer ran at only 250 lpm as opposed to the 9300's 600 lpm because
it only had half as many hammers and needed two passes of the print
bar to generate a line of print. But other changes were electronic.
The 9300's memory had a cycle time of 600 nanoseconds. The 9200
inserted a rest cycle between each active memory cycle, increasing
the effective cycle time to 1200 nanoseconds. However, there was
a switch inside labeled "Test mode D"; if you threw it those rest
cycles were eliminated and the CPU would run at full speed.
However, it also disabled parity checking, so you used it at
your own risk. The slower cycle time did allow the use of
marginal memory that wouldn't work in a 9300. You couldn't
do this if you attached disks to the machine, since the disk
channel had to magically speed up the memory to keep up with
the data flowing to and from the disk.
Post by Lew Pitcher
Apparently, it is becoming standard practice again. But, this time
in the field of consumer goods.
I asked about the rumour because I just finished reading an article
(https://www.techdirt.com/2023/05/05/mercedes-locks-better-ev-engine-performance-behind-annoying-subscription-paywalls)
about Mercedes planning to put "better engine performance" of their
EVs behind a subscription paywall.
You buy the car. If you want better engine performance, you pay
a subscription, and they /disable/ the "poorer engine performance"
component. In other words, the car that you purchased /already/ has
"better engine performance", and you pay a monthly ransom to get it
enabled.
The analogy struck me as apropos.
I heard recently that if you want to activate the heated seats
in a new BMW you have to sign up for a subscription.
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Ahem A Rivet's Shot
2023-05-07 06:51:08 UTC
Permalink
On Sun, 07 May 2023 05:02:34 GMT
Post by Charlie Gibbs
I heard recently that if you want to activate the heated seats
in a new BMW you have to sign up for a subscription.
Every large (listed) business wants to move their customers to a
subscription model because it makes growth easier and income more
predictable and Wall St. investors like you.

Some bright spark worked out that if you sell things and you want
to grow at (say) 10% then every year you have to do 110% of the selling you
did last year but if you sell subscriptions and you want to grow 10% then
every year you have only to find enough new customers to replace those
you've lost and another 10% which is much easier - especially if you're
good at keeping customers. This is now doctrine in the board rooms of
listed companies.

This is also why SAAS exists.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Freddy1X
2023-05-07 01:01:28 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
Back in the 'good old days' of PC clones, there was a switch on the front
panel to select the faster or slower processor speed supposodly because some
programs needed a certain speed for correct operation. These often came
with a jumper settable front panel display to show the current clock speed.
The ones I usually selected were 30 or 60 MHZ i think.

Now the cheapest speed 'upgrade' you could make was tho set the display
jumpers to, oh, say 100 MHZ, then you could brag to your friends.

Freddy,
...Nah! That would never happen!
--
Accept no substitutes.

/|>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>\|
/| I may be demented \|
/| but I'm not crazy! \|
/|<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<\|
* SPAyM trap: there is no X in my address *
Bob Eager
2023-05-07 10:47:11 UTC
Permalink
Post by Freddy1X
Back in the 'good old days' of PC clones, there was a switch on the
front panel to select the faster or slower processor speed supposodly
because some programs needed a certain speed for correct operation.
These often came with a jumper settable front panel display to show the
current clock speed. The ones I usually selected were 30 or 60 MHZ i
think.
I did this on my Taiwanese clone in 1985.

I had to add a small switch on the front pane;, connected to an unmarked
jumper. This increased the clock speed from 4.77MHz to 8 MHz.

I got a further speed-up by using an NEC V20 instead of the 8088. This
benchmarked very well, and it was faster, but the benchmark made it look
better than it was in everyday use. That was because one of the main
improvements was a barrel shifter, which speeded up multiplication etc.
but didn't help much with most work.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Ahem A Rivet's Shot
2023-05-07 12:07:39 UTC
Permalink
On 7 May 2023 10:47:11 GMT
Post by Bob Eager
I got a further speed-up by using an NEC V20 instead of the 8088. This
benchmarked very well, and it was faster, but the benchmark made it look
better than it was in everyday use. That was because one of the main
improvements was a barrel shifter, which speeded up multiplication etc.
but didn't help much with most work.
Om of the most impressive clock hacks I recall was a rather
ingenious one for the 80286. Starting with a 12MHz product overclocked to
the usual 15Mhz they added a circuit that bumped the clock to 40MHz until
there was a bus cycle and then dropped it to 15MHz to keep the bus happy.

The resultant AT clones dramatically outperformed the 16MHz 80386
machines that were hitting the market at the time.

Also fun was Sinclair's clock stretch hackery to sync a Z80's
irregular timing to a video clock. I never did find out who came up with
that one.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Carlos E.R.
2023-05-09 20:17:56 UTC
Permalink
Post by Ahem A Rivet's Shot
On 7 May 2023 10:47:11 GMT
Post by Bob Eager
I got a further speed-up by using an NEC V20 instead of the 8088. This
benchmarked very well, and it was faster, but the benchmark made it look
better than it was in everyday use. That was because one of the main
improvements was a barrel shifter, which speeded up multiplication etc.
but didn't help much with most work.
Om of the most impressive clock hacks I recall was a rather
ingenious one for the 80286. Starting with a 12MHz product overclocked to
the usual 15Mhz they added a circuit that bumped the clock to 40MHz until
there was a bus cycle and then dropped it to 15MHz to keep the bus happy.
:-o
Post by Ahem A Rivet's Shot
The resultant AT clones dramatically outperformed the 16MHz 80386
machines that were hitting the market at the time.
Also fun was Sinclair's clock stretch hackery to sync a Z80's
irregular timing to a video clock. I never did find out who came up with
that one.
The IBM PC clock was also related to the video timing, IIRC.
--
Cheers, Carlos.
Charlie Gibbs
2023-05-09 22:02:36 UTC
Permalink
Post by Carlos E.R.
Post by Ahem A Rivet's Shot
Also fun was Sinclair's clock stretch hackery to sync a Z80's
irregular timing to a video clock. I never did find out who came up with
that one.
The IBM PC clock was also related to the video timing, IIRC.
The Amiga was designed with video in mind. Its main clock ran
at 7.16 MHz, exactly twice the 3.58-MHz colour burst frequency.
This made it easy to build video hardware (at least for NTSC systems -
I'm not sure whether Amigas in the UK ran at a frequency better
suited to PAL). Amigas were used in a number of cable TV stations
(ever seen a Guru Meditation on TV?), and Todd Rundgren used a bank
of ten A2000s to produce his video "Change Myself".
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Carlos E.R.
2023-05-09 20:15:43 UTC
Permalink
Post by Freddy1X
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
Back in the 'good old days' of PC clones, there was a switch on the front
panel to select the faster or slower processor speed supposodly because some
programs needed a certain speed for correct operation. These often came
with a jumper settable front panel display to show the current clock speed.
The ones I usually selected were 30 or 60 MHZ i think.
Oh, it was absolutely true. Some software did not run well in faster
computers.

One example was games, like "The ancient art of war" or "The Ancient Art
of War at Sea". On a fast computer it was unplayable.

In the mid 90's, I met data acquisition and analysis software that on a
faster computer it would run accquire too many samples per second
increasing the noise. It had to run a 10 samples per second.

The root cause in both is that they used the code speed as the basis for
the timer, instead of using an actual clock or interrupts.
Post by Freddy1X
Now the cheapest speed 'upgrade' you could make was tho set the display
jumpers to, oh, say 100 MHZ, then you could brag to your friends.
Yeah, I saw that. :-)
Post by Freddy1X
Freddy,
...Nah! That would never happen!
--
Cheers, Carlos.
Ahem A Rivet's Shot
2023-05-09 20:52:44 UTC
Permalink
On Tue, 9 May 2023 22:15:43 +0200
Post by Carlos E.R.
Oh, it was absolutely true. Some software did not run well in faster
computers.
Not just DOS PCs - Xroach became impossibly fast a long time ago.
--
Steve O'Hara-Smith
Odds and Ends at http://www.sohara.org/
Mike Spencer
2023-05-09 21:45:40 UTC
Permalink
Post by Ahem A Rivet's Shot
On Tue, 9 May 2023 22:15:43 +0200
Post by Carlos E.R.
Oh, it was absolutely true. Some software did not run well in faster
computers.
Not just DOS PCs - Xroach became impossibly fast a long time ago.
xroach -speed 1 is still kinda amusing. xroach -speed 0 causes the
roaches to behavie idividually as if in the last stages of having been
sprayed with a fast-acting insecticide. Does make -squish easier to
use though. :-o

My little text editor that accepts text by displaying and punching
IBM cards works fine except that the falling chad falls too quickly to
be seen. I should fix that. RSN.
--
Mike Spencer Nova Scotia, Canada
Jeff Jonas
2023-06-09 08:42:31 UTC
Permalink
Post by Freddy1X
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds.
There were intentionally slowed-down models of the IBM 1130
but there was one contradiction:
interrupt processing required full speed to meet the time requirement.
So a workaround was to simply leave a "hanging" unfulfilled interrupt.
Post by Freddy1X
Back in the 'good old days' of PC clones, there was a switch on the front
panel to select the faster or slower processor speed
The "turbo switch" went from the standard/slow speed
(required by some games for timing loops)
to the CPU's highest speed.
Early cases simply had an LED for turbo on/off.
Later cases had a 2 digit, then a 3 digit LED display
that was merely set by the switch.
Some thought it was a speedometer and felt obligated
to upgrade their case for the 3 digit display :-0
A simple workaround was to jumper the 2 digits for "LO" and "HI".

--
James Dow Allen
2023-06-09 16:37:32 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds.
This was the case for NatSemi's lookalike version of the IBM 158, circa 1977.
originally designed by Exsysco. I don't remember the relevant model numbers.

Farfetched? I was told this by F.E.'s who claimed to have
moved the necessary wire. (Some may have been entrepreneurs
who moved the wire unbeknownst to NatSemi, pocketing
the wire's cost! :-)

Cheers,
James
Bob Eager
2023-06-09 17:20:33 UTC
Permalink
Post by James Dow Allen
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds.
This was the case for NatSemi's lookalike version of the IBM 158, circa 1977.
originally designed by Exsysco. I don't remember the relevant model numbers.
I have just been researching the ICL 2900 mainframe (not that well known).

Many models were planned but never built, or cancelled. There was a model
called the P2L (marketed as the 2960), and I became curious as to what a
P2S was. It would have been marketed as a 2950 (L ane S meant Large and
Small).

I discovered that the two would have used the same microcode engine, and
probably the same memory and other hardware. It would have been slower,
probably by using a different microcode incorporating delays.

The P2S never happened; I suspect that the P2L was slow enough, as it
didn't sell that well. The P1 and P0 never happened either.

(for completeness, P3 and P4 were each radically different, and P5 never
happened)
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Vir Campestris
2023-06-10 14:43:19 UTC
Permalink
Post by Bob Eager
I have just been researching the ICL 2900 mainframe (not that well known).
Many models were planned but never built, or cancelled. There was a model
called the P2L (marketed as the 2960), and I became curious as to what a
P2S was. It would have been marketed as a 2950 (L ane S meant Large and
Small).
I discovered that the two would have used the same microcode engine, and
probably the same memory and other hardware. It would have been slower,
probably by using a different microcode incorporating delays.
The P2S never happened; I suspect that the P2L was slow enough, as it
didn't sell that well. The P1 and P0 never happened either.
(for completeness, P3 and P4 were each radically different, and P5 never
happened)
All these years later I can't remember the internal names, but the 2950
was a real product. We had one. As we _were_ ICL that's not a complete
guide, but google says we weren't alone.

The Wikipedia article has P2S and S1 BOTH 2950!

https://en.wikipedia.org/wiki/ICL_2900_Series

It also doesn't mention the 2982 (an uprated 2980). All these years
later I can't recall the models well enough to fix Wikipedia.

Andy
Bob Eager
2023-06-10 16:28:48 UTC
Permalink
Post by Vir Campestris
Post by Bob Eager
I have just been researching the ICL 2900 mainframe (not that well known).
Many models were planned but never built, or cancelled. There was a
model called the P2L (marketed as the 2960), and I became curious as to
what a P2S was. It would have been marketed as a 2950 (L ane S meant
Large and Small).
I discovered that the two would have used the same microcode engine,
and probably the same memory and other hardware. It would have been
slower, probably by using a different microcode incorporating delays.
The P2S never happened; I suspect that the P2L was slow enough, as it
didn't sell that well. The P1 and P0 never happened either.
(for completeness, P3 and P4 were each radically different, and P5
never happened)
All these years later I can't remember the internal names, but the 2950
was a real product. We had one. As we _were_ ICL that's not a complete
guide, but google says we weren't alone.
The Wikipedia article has P2S and S1 BOTH 2950!
https://en.wikipedia.org/wiki/ICL_2900_Series
It also doesn't mention the 2982 (an uprated 2980). All these years
later I can't recall the models well enough to fix Wikipedia.
Sorry, I should have been clearer. The P2S was *going* to be the 2950.
However, as I said it didn't fit well in the range (and P series were
expensive).

They still had a 2950, but it wasn't a P2S. The name was attached to the
S1L, announced in November 1977. I think slightly more powerful, but
smaller and cheaper to produce.

That Wikipedia article is short, but the diagram for Modules aand
Interconnections is woefully wrong. I did do a correct one for a talk I
gave recently, but I can't deal with the hassle of getting Wikipedia
corrected.

(I managed a P2L (later, as a dual) for about 8 years)
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Bob Eager
2023-06-10 16:32:18 UTC
Permalink
Post by Vir Campestris
It also doesn't mention the 2982 (an uprated 2980). All these years
later I can't recall the models well enough to fix Wikipedia.
I'm not convinced the 2982 (probably a P4L) was ever delivered, although
it's mentioned. I think the same thing happened as with the 2950, but they
gave it a different name. It would have been an S series machine, probably
an S3 variant (which is all ICL made in the end), and named the 2988
(which did exist).
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Vir Campestris
2023-06-12 14:59:14 UTC
Permalink
Post by Bob Eager
Post by Vir Campestris
It also doesn't mention the 2982 (an uprated 2980). All these years
later I can't recall the models well enough to fix Wikipedia.
I'm not convinced the 2982 (probably a P4L) was ever delivered, although
it's mentioned. I think the same thing happened as with the 2950, but they
gave it a different name. It would have been an S series machine, probably
an S3 variant (which is all ICL made in the end), and named the 2988
(which did exist).
Again, I was internal, so it's no guarantee of what customers saw - but
ICL Bracknell had a 2980 which was upgraded to a 2982. IIRC that was
basically the same machine, but clocked a bit faster. It was definitely
the same family.

Whether any were _sold_ is another matter.

Andy
Bob Eager
2023-06-12 20:52:00 UTC
Permalink
Post by Vir Campestris
All these years later I can't remember the internal names, but the 2950
was a real product. We had one. As we _were_ ICL that's not a complete
guide, but google says we weren't alone.
If you are still interested after all these years, I recently gave a
lecture on a 2900 operating system (not VME):

http://emas.bobeager.uk

(link on that page)
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Andy Burns
2023-06-13 02:01:37 UTC
Permalink
Post by Bob Eager
If you are still interested after all these years, I recently gave a
http://emas.bobeager.uk
(link on that page)
authentication required?
Bob Eager
2023-06-13 07:47:27 UTC
Permalink
Post by Andy Burns
Post by Bob Eager
If you are still interested after all these years, I recently gave a
http://emas.bobeager.uk
(link on that page)
authentication required?
It should be OK except for the 2900 series page, which is under
development. The video on the page I mentioned is the one.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Vir Campestris
2023-06-18 11:09:18 UTC
Permalink
Post by Bob Eager
Post by Andy Burns
Post by Bob Eager
If you are still interested after all these years, I recently gave a
http://emas.bobeager.uk
(link on that page)
authentication required?
It should be OK except for the 2900 series page, which is under
development. The video on the page I mentioned is the one.
I read the slides. Interesting!

Bob, one thing I hate you for. My computer lives in my garden office,
and I have a couple of dead pixels on the screen. Actually not pixels,
they're dead thrips (small bugs) which have crawled in and died.

So when I saw that bug running around in the corner of your web page I
had a moment's panic!

Your summary of VME/K is quote interesting. As you say, ICL started a
project to re-architect it, just as you dropped it. That project
improved the system reliability enormously. Just as it was getting to a
decent standard ICL decided to can the whole OS!

Andy
Bob Eager
2023-06-18 13:37:43 UTC
Permalink
Post by Vir Campestris
Post by Bob Eager
Post by Andy Burns
Post by Bob Eager
If you are still interested after all these years, I recently gave a
http://emas.bobeager.uk
(link on that page)
authentication required?
It should be OK except for the 2900 series page, which is under
development. The video on the page I mentioned is the one.
I read the slides. Interesting!
Bob, one thing I hate you for. My computer lives in my garden office,
and I have a couple of dead pixels on the screen. Actually not pixels,
they're dead thrips (small bugs) which have crawled in and died.
So when I saw that bug running around in the corner of your web page I
had a moment's panic!
Your summary of VME/K is quote interesting. As you say, ICL started a
project to re-architect it, just as you dropped it. That project
improved the system reliability enormously. Just as it was getting to a
decent standard ICL decided to can the whole OS!
At that point I think they were spending 35% of their money on R&D, and Ed
Mack was not the most popular guy, I believe.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Bill Findlay
2023-06-18 23:32:59 UTC
Permalink
Post by Bob Eager
Post by Vir Campestris
Post by Bob Eager
Post by Andy Burns
Post by Bob Eager
If you are still interested after all these years, I recently gave a
http://emas.bobeager.uk
(link on that page)
authentication required?
It should be OK except for the 2900 series page, which is under
development. The video on the page I mentioned is the one.
I read the slides. Interesting!
Bob, one thing I hate you for. My computer lives in my garden office,
and I have a couple of dead pixels on the screen. Actually not pixels,
they're dead thrips (small bugs) which have crawled in and died.
So when I saw that bug running around in the corner of your web page I
had a moment's panic!
Your summary of VME/K is quote interesting. As you say, ICL started a
project to re-architect it, just as you dropped it. That project
improved the system reliability enormously. Just as it was getting to a
decent standard ICL decided to can the whole OS!
At that point I think they were spending 35% of their money on R&D, and Ed
Mack was not the most popular guy, I believe.
He should never have been allowed to have his own private OS,
diverting scarce resources away from other products.
--
Bill Findlay
Bob Eager
2023-06-19 08:41:16 UTC
Permalink
On 18 Jun 2023, Bob Eager wrote (in article
Post by Bob Eager
Post by Vir Campestris
Post by Bob Eager
Post by Andy Burns
Post by Bob Eager
If you are still interested after all these years, I recently
http://emas.bobeager.uk
(link on that page)
authentication required?
It should be OK except for the 2900 series page, which is under
development. The video on the page I mentioned is the one.
I read the slides. Interesting!
Bob, one thing I hate you for. My computer lives in my garden office,
and I have a couple of dead pixels on the screen. Actually not
pixels, they're dead thrips (small bugs) which have crawled in and
died.
So when I saw that bug running around in the corner of your web page
I had a moment's panic!
Your summary of VME/K is quote interesting. As you say, ICL started a
project to re-architect it, just as you dropped it. That project
improved the system reliability enormously. Just as it was getting to
a decent standard ICL decided to can the whole OS!
At that point I think they were spending 35% of their money on R&D, and
Ed Mack was not the most popular guy, I believe.
He should never have been allowed to have his own private OS,
diverting scarce resources away from other products.
The trouble was that the P2 (2960) and the projected 2940 (P1) and 2930
(P0) were too weak for VME/B at the time. Even the revised 2930 (S0)
wouldn't have worked.

But Mack certainly had too much power.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
Vir Campestris
2023-06-20 11:38:45 UTC
Permalink
Post by Bob Eager
The trouble was that the P2 (2960) and the projected 2940 (P1) and 2930
(P0) were too weak for VME/B at the time. Even the revised 2930 (S0)
wouldn't have worked.
The rumour I heard was than when they put "VME/2900" (renamed VME/B) on
the K sites it was supposed to need 1MB of extra RAM, and that was all.

The couldn't switch all sites to K because of a few odd things like DAPs.

And it turned out to be double the RAM (which was 1MB on most sites) but
more critically an extra 10% of CPU.

The rumour also said that it was somebody from the B project who wrote
the report!
Post by Bob Eager
But Mack certainly had too much power.
I know the name, but I think he might predate me.

Andy
Bob Eager
2023-06-20 13:06:21 UTC
Permalink
Post by Vir Campestris
Post by Bob Eager
The trouble was that the P2 (2960) and the projected 2940 (P1) and 2930
(P0) were too weak for VME/B at the time. Even the revised 2930 (S0)
wouldn't have worked.
The rumour I heard was than when they put "VME/2900" (renamed VME/B) on
the K sites it was supposed to need 1MB of extra RAM, and that was all.
I think by that time it was considerably faster. The early tests at
Edinburgh showed it supporting a tiny MAC load on a 2970. God knows what
it would have been like on a 2960.

As it was, ICL gave us another 2MB of RAM, and we sourced a second OCP
(but if you've read the slides, I think they said that).
Post by Vir Campestris
The couldn't switch all sites to K because of a few odd things like DAPs.
Indeed. The DAP was quite good; they had some at Edinburgh, on EMAS.
Post by Vir Campestris
Post by Bob Eager
But Mack certainly had too much power.
I know the name, but I think he might predate me.
Ed Mack was ejected at the same time that VME/K was killed.
--
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
http://www.mirrorservice.org
gareth evans
2023-06-20 15:08:30 UTC
Permalink
Post by Vir Campestris
Post by Bob Eager
The trouble was that the P2 (2960) and the projected 2940 (P1) and 2930
(P0) were too weak for VME/B at the time. Even the revised 2930 (S0)
wouldn't have worked.
The rumour I heard was than when they put "VME/2900" (renamed VME/B) on
the K sites it was supposed to need 1MB of extra RAM, and that was all.
The couldn't switch all sites to K because of a few odd things like DAPs.
And it turned out to be double the RAM (which was 1MB on most sites) but
more critically an extra 10% of CPU.
the report!
Post by Bob Eager
But Mack certainly had too much power.
I know the name, but I think he might predate me.
The urban legend was that 1900 was the year that the hardware
was designed and that 2900 is he year that the software will be ready.

1970 - 71 2nd year electronics at Essex, the PDP8 and the 1900 were the
two examples of computer design that we studied.

ISTR the textbook Digital Computer Design by F G Heath described the
architecture of the 1900.

Charlie Gibbs
2023-06-09 18:24:41 UTC
Permalink
Post by James Dow Allen
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds.
This was the case for NatSemi's lookalike version of the IBM 158, circa 1977.
originally designed by Exsysco. I don't remember the relevant model numbers.
The Univac 9200, a de-rated version of the 9300 (their answer
to the IBM 360/20), inserted a "rest cycle" between each real
memory cycle, increasing the memory's effective cycle time from
600 nanoseconds to 1200 ns. But that made it too slow to handle
those newfangled disk drives - so the selector channel would do
some magic to eliminate the rest cycles while it was active.
This meant that if your machine had marginal memory that couldn't
handle the 600-ns cycle time, you'd have to replace it as part of
adding disks to the system.
Post by James Dow Allen
Farfetched? I was told this by F.E.'s who claimed to have
moved the necessary wire. (Some may have been entrepreneurs
who moved the wire unbeknownst to NatSemi, pocketing
the wire's cost! :-)
Inside the 9200 was a secret toggle switch labeled TEST MODE D.
(There were test mode A, B, and C switches on the front panel.)
If you threw this switch, memory would run at full speed.
However, parity checking was also disabled, so you used it
at your own risk. Normally, if a parity error occurred, the
processor would freeze and the PROC ABN (processor abnormal)
light would come on. One day we were running a large sort job
with test mode D set. A parity error occurred, and I was treated
to the sight of the machine furiously churning away with the
PROC ABN light on. Needless to say, we had to cancel the job
and re-run it.
--
/~\ Charlie Gibbs | You can't save the earth
\ / <***@kltpzyxm.invalid> | unless you're willing to
X I'm really at ac.dekanfrus | make other people sacrifice.
/ \ if you read it the right way. | -- Dogbert the green consultant
Scott Lurndal
2023-05-07 16:10:24 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
Converting from a Burroughs B4925 to a B4955 was a jumper on the fetch
board to enable dual-issue and a firmware update.
s***@alum.dartmouth.org
2023-05-07 18:22:32 UTC
Permalink
The difference between a GE-255 and a 265 was a delay loop in the OS.

In alt.folklore.computers Lew Pitcher <***@digitalfreehold.ca> wrote:
: Long ago, I was told (or I read) about an early business computer
: (I'm guessing circa 1960) that the manufacturer sold in two
: different clock speeds. If you bought the "slower" system, you
: could pay a licence fee and have a CSR come to /remove/ a component
: that /limited/ the CPU speed.

: Does any one else know of the details of this? Was it a rumour,
: or did it actually happen?

: --
: Lew Pitcher
: "In Skills We Trust"
Alfred Falk
2023-05-07 23:32:23 UTC
Permalink
Post by Lew Pitcher
Long ago, I was told (or I read) about an early business computer
(I'm guessing circa 1960) that the manufacturer sold in two
different clock speeds. If you bought the "slower" system, you
could pay a licence fee and have a CSR come to /remove/ a component
that /limited/ the CPU speed.
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
CDC Cyber 72 was an example. From https://en.wikipedia.org/wiki/CDC_Cyber:
" The Cyber-72 had identical hardware to a Cyber-73, but added additional
clock cycles to each instruction to slow it down. This allowed CDC to offer
a lower performance version at a lower price point without the need to
develop new hardware."

I was told by someone who had worked at CDC, that slower model actually cost
about %65 more to make. Another aspect of the lower cost was that the
slower cycle time allowed use of "reject" memory that did not perform
reliably at full speed.
There are lots of other examples of deliberate performance throttling for
marketing purposes, but also to allow sales of "reject" units. Intel's
binning of CPU's are an example.
I once worked on a Wang minicomputer that had a (IIRC) 20 MB disk that was
sold in 5, 10, 15, and 20 MB versions. Field upgrade to a larger size
involved inserting a small circuit board (a simple jumpter would be to easy
cheat) that enabled larger address range.
Vir Campestris
2023-05-11 16:51:06 UTC
Permalink
Post by Lew Pitcher
Does any one else know of the details of this? Was it a rumour,
or did it actually happen?
I too have heard the rumour. The detail I heard was that a field
engineer would come out and spend a day (1) moving the jumper and (2)
testing it all to make sure it worked.

They always did work, but a day of an on site engineer looked better ;)

Andy
Loading...