6 Aug
2012
6 Aug
'12
11:59 p.m.
After looking at nice computers at Ames and the Computer History Museum, a
though occurred to me: What is the slowest/oldest supercomputer still in
operation?
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
7 Aug
7 Aug
6:40 a.m.
oldest I have seen run in a production environment is David Smiths IBM
System 3 at Dyserv in Columbus Ohio.
The other Bob
-----Original Message-----
From: David Griffith
Sent: Tuesday, August 07, 2012 2:59 AM
To: cctalk at classiccmp.org
Subject: slowest supercomputer still in operation
After looking at nice computers at Ames and the Computer History Museum, a
though occurred to me: What is the slowest/oldest supercomputer still in
operation?
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
8:31 a.m.
Now be fair, that wasn't considered a supercomputer then or now. Of course here's
where semantics come in. Accepting that Stretch was the first, and the first surviving is
likely to be a Cray1, who has one working? TNMOC has a 90s air cooled Cray model which is
operational, and an Altrix (not on display), but surely that's not the oldest. There
are a number of mainstream early systems like PDPs, 1401s, the Pegasus, etc. are out
there but they don't count. How about a large CDC or IBM?
Bob Bradlee <caveguy at sbcglobal.net> wrote:
oldest I have seen run in a production environment is
David Smiths IBM
System 3 at Dyserv in Columbus Ohio.
The other Bob
-----Original Message-----
From: David Griffith
Sent: Tuesday, August 07, 2012 2:59 AM
To: cctalk at classiccmp.org
Subject: slowest supercomputer still in operation
After looking at nice computers at Ames and the Computer History
Museum, a
though occurred to me: What is the slowest/oldest supercomputer still
in
operation?
--
David Griffith
dgriffi at cs.csubak.edu
A: Because it fouls the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
-- Colin
10:02 a.m.
On 7 Aug 2012 at 16:31, Colin Eby wrote:
Now be fair, that wasn't considered a
supercomputer then or now. Of
course here's where semantics come in. Accepting that Stretch was the
first, and the first surviving is likely to be a Cray1, who has one
working? TNMOC has a 90s air cooled Cray model which is operational,
and an Altrix (not on display), but surely that's not the oldest.
There are a number of mainstream early systems like PDPs, 1401s, the
Pegasus, etc. are out there but they don't count. How about a large
CDC or IBM?
Is the CM-1 at CHM functional?
--Chuck
11:54 a.m.
On 7 Aug 2012 at 10:33, Fred Cisin wrote:
Does functional and fired up as demonstrations count
as "still in
operation"?
How is functionality assured if there's no software? I mean, you've
got a bunch of red LEDs...
Does anyone have an operational ETA-10P ("Piper")? I think that
might qualify as a "supercomputer"... Convex also had some low-end
models, but I don't recall the model numbers.
And I'm sure there are some Sequent systems kicking around, but that
may be stretching the "supercomputer" definition a bit.
--Chuck
10:40 a.m.
From: cclist at sydex.com
On 7 Aug 2012 at 16:31, Colin Eby wrote:
Now be fair, that wasn't considered a
supercomputer then or now. Of
course here's where semantics come in. Accepting that Stretch was the
first, and the first surviving is likely to be a Cray1, who has one
working? TNMOC has a 90s air cooled Cray model which is operational,
and an Altrix (not on display), but surely that's not the oldest.
There are a number of mainstream early systems like PDPs, 1401s, the
Pegasus, etc. are out there but they don't count. How about a large
CDC or IBM?
Is the CM-1 at CHM functional?
--Chuck Hi I doubt it. From what I heard, no one has any softwareto run on it, even if
powered up.Dwight
4:51 p.m.
On 7 Aug 2012 at 16:06, William Donzelli wrote:
I don't know that I'd call the 180-960 a "supercomputer". It's much
later (1988) nd not as fast as a 7600 (1969)
The Cray-Cyber folks clain a Cray YMP-EL, which, again, is somewhat
questionable as "supercomputer" (1992, 133Mflop/CPU) considering the
time. Compare, for example, with a "real" supercomputer from 5 years
earlier, the ETA-10 running at 10Gflop.
"Supercomputer" to my mind, depends on a the relative level of
performance for the time.
--Chuck
How about a
large CDC or IBM?
There are still Cyber 960s (in very small amounts). The CDC water
machines are out of service. 
6:38 p.m.
On 08/07/2012 07:51 PM, Chuck Guzis wrote:
I don't know that I'd call the 180-960 a "supercomputer". It's
much
later (1988) nd not as fast as a 7600 (1969)
The Cray-Cyber folks clain a Cray YMP-EL, which, again, is somewhat
questionable as "supercomputer" (1992, 133Mflop/CPU) considering the
time. Compare, for example, with a "real" supercomputer from 5 years
earlier, the ETA-10 running at 10Gflop.
"Supercomputer" to my mind, depends on a the relative level of
performance for the time.
I have several YMP-ELs. This is a muddy area...They were sold as
low-cost (~$300K) development machines for development, debugging, and
generation of executables for "big" YMP-class supercomputers. As such,
they are binary-compatible.
It's faster than a Cray-1, and they're aesthetically very pretty. (a
hallmark of supercomputers! ;))
Most people (myself included) would consider it a supercomputer. When
it was sold, it was considered a supercomputer. There are nearly always
"faster" (and what constitutes "faster"?) computers out there. Does
that mean that a given machine is somehow not a "supercomputer", because
there's another machine out there that's faster?
MFLOP/S per dollar? Per watt? Per pound? Per cubic foot? Per
coolant loop? Vector length? Vector PRESENCE? Parallelism? But then
what is parallelism? (parallel bits, parallel instructions, parallel
processors?)
This whole obsession with strict definitions and classification, which
we slide into with some frequency on this list, is rather silly, don't
you think?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
How about a large CDC or IBM?
There are still Cyber 960s (in very small amounts). The CDC water
machines are out of service.
7:41 p.m.
On 7 Aug 2012 at 21:38, Dave McGuire wrote:
MFLOP/S per dollar? Per watt? Per pound? Per
cubic foot? Per
coolant loop? Vector length? Vector PRESENCE? Parallelism? But
then what is parallelism? (parallel bits, parallel instructions,
parallel processors?)
Allow me to propose a measure--LINPACK. Should run on just about
everything and with excellent pedigree. Further, to make it time-
sensitive, let's define a supercomputer as a machine that has a
LINPACK score at least 75% of that of the most powerful machine
offered by any vendor in the subject machine's year of introduction.
And AFAIK, all supercomputers eat FORTRAN.
The game with supercomputers is to blow away the competition, cost
being very secondary. "Budget" versions with substantially lower
performance don't count. 75% will accommodate near-misses.
So, an ETA-10P doesn't qualify, but an ETA-10 does for 1988. Binary-
compatible all.
A STAR-1B or STAR-65 doesn't qualify, but a STAR-100 does for 1970.
Ditto.
A YMP-EL was not the state of the art for supercomputers in 1992.
133Mflops wasn't state-of-the-art in 1982. Maybe 1972.
Bring it on...
--Chuck
9:39 p.m.
On 7 Aug 2012 at 22:51, William Donzelli wrote:
Why not? The scores are well documented going back to 1978. The
2011 report:
ftp://www.netlib.org/benchmark/performance.pdf
Linpack was significant enough that a computer manufacturer tooks its
name from one of the tests (the idea was that their architecture was
peculiarly suited to the characteristics of the test).
--Chuck
Bring it on...
"Page one of the LINPACK report"
8:36 a.m.
On 7 Aug 2012 at 9:40, Bob Bradlee wrote:
oldest I have seen run in a production environment is
David Smiths
IBM System 3 at Dyserv in Columbus Ohio.
I'd hardly call a System/3 a "supercomputer". Given the operating
and maintenance headaces and the demise of vendors, I suspect the
oldest operating one might a massively parallel one from the 1990s.
But I'm not sure.
--Chuck
1:29 p.m.
On 7 Aug 2012 at 15:58, William Donzelli wrote:
Well, so did the 1401 and 1620. Supercomputers only if you were on a
controlled substance, I guess. Uniprocessor supercomputers had the
problem that they were very much bleeding-edge affairs, with little
thought given to maintenance or resource consumption. Parts were
often custom fab, assemblies could be hell on wheels to service.
Wasn't the MTTR for "Bubbles" something like 8 hours? I know it was
very high for the ETA-10 also.
I suspect the number of still-in-use 80s uniprocessor supercomputers
would probably be a close match to the number of extant Saxpy Matrix-
1s.
--Chuck
I'd hardly
call a System/3 a "supercomputer".
It did have FORTRAN! I wonder what kind of a trainwreck that was. 
7:56 p.m.
> I'd hardly call a System/3 a
"supercomputer".
On Tue, 7 Aug 2012, William Donzelli wrote:
It did have FORTRAN! I wonder what kind of a
trainwreck that was.
FORTRAN hardly a "supercomputer" makes!
TRS80 Model 1 had FORTRAN! I personally sold the first three retail
copies! Microsoft had delayed it repeatedly; I was passing through the
Seattle airport, and Bob Wallace forced that department's hand and got me
three copies delivered straight from Microsoft to me at the Seattle
airport.
5150 had FORTRAN. Now THAT was a trainwreck! On simple benchmarks, such
as an integer Sieve of Erastothanes, it was slower than the 5150
interpreted BASIC!
9:45 p.m.
On 7 Aug 2012 at 19:56, Fred Cisin wrote:
5150 had FORTRAN. Now THAT was a trainwreck! On
simple benchmarks,
such as an integer Sieve of Erastothanes, it was slower than the 5150
interpreted BASIC!
Do you recall who did the FORTRAN? Was it Ryan-McFarland? If so, I
can understand the speed.
Microsoft F80 for 8080 had pretty decent code generation. I think
the FORTRAN for ISIS-II (Intel MDS) came on something like 7 8"
floppies. It was BIG.
Was there a FORTRAN for the 360/20?
--Chuck
8 Aug
8 Aug
11:53 a.m.
> 5150 had FORTRAN. Now THAT was a trainwreck! On
simple benchmarks,
> such as an integer Sieve of Erastothanes, it was slower than the 5150
> interpreted BASIC!
On Tue, 7 Aug 2012, Chuck Guzis wrote:
Do you recall who did the FORTRAN? Was it
Ryan-McFarland? If so, I
can understand the speed.
I'm referring to Microsoft FORTRAN for the 5150, as sold by IBM in 1983?
It was written in MS-Pascal, (Bob Wallace wrote the MS-Pascal). and I was
advised to try to avoid using its run-time library. Variable names of 6?
characters long, ending in 'QQ' were off-limits.
Nevertheless, in spite of poor performance, it did just fine for its
intended use, which was to TEACH FPORTRAN
Microsoft F80 for 8080 had pretty decent code
generation. I think
the FORTRAN for ISIS-II (Intel MDS) came on something like 7 8"
floppies. It was BIG.
MS-FORTRAN for the 5150 was a new re-write, not a port. (Or at least that
was what I was told).
MS-FORTRAN for the TRS80 was actually decent, and IIRC, only a couple of
disks, although the manual had a bunch of typos. An instructor at the
community college where I sold the first copy insisted that the compiler
was "hopelessly flawed", because he typed in the first demo program in the
manual and it generated a syntax error due to a typo. He was
well-regarded by his colleagues, and displeased when I suggested that a
TEACHER of FORTRAN should surely be ably to debug such an obvious error,
for his was too highly evolved a mind to concern itself with whether it
should be a comma or a full-stop between the device number and the FORMAT
statement number in a WRITE statement (such being the ways of the Grand
Academy of Lagado, and most of the rest of Academia). When he moved to a
different university a few years later, I was given his job, starting with
teaching FORTRAN, now on 5150 with MS-FORTRAN.
45 years ago, on the 1620, we had PDQ ("Pretty Damn Quick" (a lie))
FORTRAN, IIRC, it was two passes to compile, each one called for a 3 or 4
inch deck of cards for thqat phase of the compiler. I had previously
written FORTRAN on 360s at GWU, but the 1620 was my first chance to
OPERATE the computer.
I never was allowed entry into the grand temple room of the 1401.
Was there a FORTRAN for the 360/20?
There was
for the 360/30 at least. Later there was WATFOR
My father did his work at GWU (George Washington University).
I did a lot of keypunching, verifying, and sorting.
Eventually I was allowed to load cards into the 360's reader, and
ultimately to load paper (I mastered the ability to load paper without
pausing the printer - much like threading a 16mm projector without
stopping). Although I got an opportunity and took a semester course in
"360 Operator Training" at the United States Department of Agriculture
Graduate School, I never was allowed to touch the console.
--
Grumpy Ol' Fred cisin at xenosoft.com
2:48 p.m.
On 8 Aug 2012 at 11:53, Fred Cisin wrote:
Well, yes--there was a bare-bones "USA BASIC FORTRAN" for the smaller
32-bit S/360s (I used it under DOS. It was distinguished by the fact
that it probably had fewer than 10 one-or-two-word error messages),
but I wonder if there was ever one for the 16-bit model 20.
--Chuck
I'm referring to Microsoft FORTRAN for the 5150,
as sold by IBM in
1983?
Just wondering--R-M had both FORTRAN and COBOL for the 5150 pretty
early on.
45 years ago, on the 1620, we had PDQ ("Pretty
Damn Quick" (a lie))
FORTRAN, IIRC, it was two passes to compile, each one called for a 3
or 4 inch deck of cards for thqat phase of the compiler. I had
previously written FORTRAN on 360s at GWU, but the 1620 was my first
chance to OPERATE the computer.
I used both the card-oriented IBM FORTRAN (read in the first pass,
then your source, cross your fingers and hope the 1622 doesn't flake
out while punching the intermediate results, the read in the second
pass, collect the punched object, then read it in with loader and
library. Lots places for things to get messed up.
Later I used FORTRAN II-D under Monitor II on the 1620 with an 1311
disk. It worked much better and you could call SPS subroutines from
it. No user-oriented file system per-se; just "work cylinders" that
were fair game for any program and the "system area" with fixed
locations for ssystem code..
Ever use it?
Was there a
FORTRAN for the 360/20?
There was for the 360/30 at least. Later there was WATFOR
My father did his work at GWU (George Washington University).
I did a lot of keypunching, verifying, and sorting.
5 p.m.
> I'm referring to Microsoft FORTRAN for the
5150, as sold by IBM in
> 1983?
On Wed, 8 Aug 2012, Chuck Guzis wrote:
Just wondering--R-M had both FORTRAN and COBOL for the
5150 pretty
early on.
But, surely not as slow as the IBM MS-DORTRAN??!?
Ever use it?
No, the only FORTRAN that I had
access to on the 1620 (as a student in
community college) was the "PDQ"
But the 1401 emulator was even more fun.
Well, yes--there was a bare-bones "USA BASIC
FORTRAN" for the smaller
32-bit S/360s (I used it under DOS. It was distinguished by the fact
that it probably had fewer than 10 one-or-two-word error messages),
but I wonder if there was ever one for the 16-bit model 20.
No idea. But the WATFOR had better error messages. Which, as an
inexperienced programmer, I needed.
7 Aug
7 Aug
1:03 p.m.
When I was cleaning out CyberResources maybe three years back, the
e-recycler he was using received the power supplies from a Cray X/MP
from Canada. He thought the rest of the machine was
still in Canada,
but doomed. It may have been an operational machine at the time.
--
Will
10:16 a.m.
ok. the challenge has been offered.... let the responses begin!
steve
Let the gOn 8/7/2012 12:54 PM, Fred Cisin wrote:
On Mon, 6 Aug 2012, David Griffith wrote:
After looking at nice computers at Ames and the
Computer History Museum, a
though occurred to me: What is the slowest/oldest supercomputer still in
operation?
What IS a "supercomputer"?
(some people define it in terms of speed!)
10:38 a.m.
do hobbyists running cray-1's count? cuz there's a few of those around...
Date: Tue, 7 Aug 2012 13:16:03 -0400
From: shumaker at att.net
To: cctalk at classiccmp.org
Subject: Re: slowest supercomputer still in operation
ok. the challenge has been offered.... let the responses begin!
steve
Let the gOn 8/7/2012 12:54 PM, Fred Cisin wrote:
On Mon, 6 Aug 2012, David Griffith wrote:
After looking at nice computers at Ames and the
Computer History Museum, a
though occurred to me: What is the slowest/oldest supercomputer still in
operation?
What IS a "supercomputer"?
(some people define it in terms of speed!)
2:06 p.m.
On 08/07/2012 11:54 AM, Fred Cisin wrote:
On Mon, 6 Aug 2012, David Griffith wrote:
Defining "still in operation" might not be a bad idea, either...
After looking at nice computers at Ames and the
Computer History Museum, a
though occurred to me: What is the slowest/oldest supercomputer still in
operation?
What IS a "supercomputer"?
(some people define it in terms of speed!)
4518
days inactive
4520
days old
29 comments
12 participants
participants (12)
-
caveguy＠sbcglobal.net -
cclist＠sydex.com -
cisin＠xenosoft.com -
colineby＠isallthat.com -
dgahling＠hotmail.com -
dgriffi＠cs.csubak.edu -
dkelvey＠hotmail.com -
eric＠brouhaha.com -
jules.richardson99＠gmail.com -
mcguire＠neurotica.com -
shumaker＠att.net -
wdonzelli＠gmail.com