A Complete History Of Mainframe Computing
Anne & Lynn Wheeler
A Complete History Of Mainframe Computing
http://www.tomshardware.com/picturestory/508-mainframe-computer-history.html
--
40+yrs virtualization experience (since Jan68), online at home since Mar1970
Quadibloc
> A Complete History Of Mainframe Computing
One of the photos was wrong...
on the page about the Burroughs B5000 computer, they have instead a
photograph of the BRLESC computer.
John Savard
Brian Boutel
> A Complete History Of Mainframe Computing
> http://www.tomshardware.com/picturestory/508-mainframe-computer-history.html
>
Hardly complete, unless the world consists entirely of North America.
--
Wellington, New Zealand
"I don't respond to Christopher Hitchens in public, on the general
principle that you should never mud-wrestle with a pig because you both
get filthy and the pig likes it." -- Tony Judt
Peter Flass
> Anne & Lynn Wheeler wrote:
>> A Complete History Of Mainframe Computing
>> http://www.tomshardware.com/picturestory/508-mainframe-computer-history.html
>>
>>
>
> Hardly complete, unless the world consists entirely of North America.
>
>
It's like that "New Yorker" map. A few places on the US East Coast are
shown very large, the rest of North America somewhat smaller, and Urrup
about the size if Iceland.
Walter Bushell
Peter Flass <Peter...@Yahoo.com> wrote:
But California is bigger than the rest of the country outside NYC,
flyover country is really shrunk.
Scott Lurndal
>Anne & Lynn Wheeler wrote:
>> A Complete History Of Mainframe Computing
>> http://www.tomshardware.com/picturestory/508-mainframe-computer-history.html
>>
>
>Hardly complete, unless the world consists entirely of North America.
>
It wasn't even complete for North America. Honeywell, GE, Electrodata,
most of the Burroughs systems (B1700, B4700, et. al.), NCR, all missing.
scott
Charlie Gibbs
(Peter Flass) writes:
Recently I went to a restaurant that still has those satirical
British Columbia placemats that I remember from at least 40 years
ago. B.C. takes up half the map of Canada, and that large country
to the south is labeled "Unexplored Southern Area".
--
/~\ cgi...@kltpzyxm.invalid (Charlie Gibbs)
\ / I'm really at ac.dekanfrus if you read it the right way.
X Top-posted messages will probably be ignored. See RFC1855.
/ \ HTML will DEFINITELY be ignored. Join the ASCII ribbon campaign!
Quadibloc
> Hardly complete, unless the world consists entirely of North America.
Any claim to a "complete" history of mainframe computing for something
that is less than a 20-volume work is, of necessity, pretentious.
Others have noted that some North American computer makers weren't
mentioned either. On the other hand, the PDP-8 was a minicomputer, not
a mainframe.
It would indeed have been nice to see mention of the KDF-9 or the ICL
1900. Or the Cray-1, or NCR mainframes... but it was a reasonable
short survey of the subject.
John Savard
CBFalconer
> Peter...@Yahoo.com (Peter Flass) writes:
>
... snip ...
>
>> It's like that "New Yorker" map. A few places on the US East
>> Coast are shown very large, the rest of North America somewhat
>> smaller, and Urrup about the size if Iceland.
>
> Recently I went to a restaurant that still has those satirical
> British Columbia placemats that I remember from at least 40
> years ago. B.C. takes up half the map of Canada, and that large
> country to the south is labeled "Unexplored Southern Area".
Very useful. They should market those (at a reasonable price).
--
[mail]: Chuck F (cbfalconer at maineline dot net)
[page]: <http://cbfalconer.home.att.net>
Try the download section.
Morten Reistad
Quadibloc <jsa...@ecn.ab.ca> wrote:
>On Jun 27, 11:50�pm, Brian Boutel <f...@fake.nz> wrote:
>
>> Hardly complete, unless the world consists entirely of North America.
>
>Any claim to a "complete" history of mainframe computing for something
>that is less than a 20-volume work is, of necessity, pretentious.
Ant the title of that work would be closer to "ein kurzer einf�hrung"
(a short introduction) than to anything comprehensive.
>Others have noted that some North American computer makers weren't
>mentioned either. On the other hand, the PDP-8 was a minicomputer, not
>a mainframe.
>
>It would indeed have been nice to see mention of the KDF-9 or the ICL
>1900. Or the Cray-1, or NCR mainframes... but it was a reasonable
>short survey of the subject.
-- mrr
Christian Brunschen
Morten Reistad <fi...@last.name> wrote:
[ snippage ]
>"ein kurzer einf�hrung"
"Eine kurze Einf�hrung"
[ snippage ]
>-- mrr
// Christian Brunschen
grey...@mail.com
> Charlie Gibbs wrote:
>> Peter...@Yahoo.com (Peter Flass) writes:
>>
> ... snip ...
>>
>>> It's like that "New Yorker" map. A few places on the US East
>>> Coast are shown very large, the rest of North America somewhat
>>> smaller, and Urrup about the size if Iceland.
>>
>> Recently I went to a restaurant that still has those satirical
>> British Columbia placemats that I remember from at least 40
>> years ago. B.C. takes up half the map of Canada, and that large
>> country to the south is labeled "Unexplored Southern Area".
>
> Very useful. They should market those (at a reasonable price).
>
Add in those German Beer ads (was it?) in Mexico that showed
California, arizona, new mexico as part of Mexico..
--
Greymaus
.
.
...
hanc...@bbs.cpcn.com
> A Complete History Of Mainframe Computinghttp://www.tomshardware.com/picturestory/508-mainframe-computer-histo...
Thanks for sharing the reference.
I would've included the IBM SSEC. While it was technologically no big
deal--some electronics but mostly relay--it still was programmable and
IBM got some valuable patents and valuable programming and operations
experience out of it. Many of the people who worked on it
subsequently worked on IBM's other mainframes and their experience was
a big contribution.
Although some might say the IBM 701 was technically inferior to the
Univac I, IIRC the IBM 701 got up and running and did useful work at
customer sites immediately after delivery. I can't help but suspect
that IBM was better at helping the customer do early programming but
also in organizing the operations environment to be efficient.
Likewise for the 702.
(The IBM history book "IBM Early Computers" goes into details on this,
though not on the Univac.)
I had business at Univac's HQ in the 1970s and it was quite obviously
a different company than IBM, very different atmosphere. IBM was very
"waspy" (regardless of the ethnicity of the person, black, green, or
whatever, everyone became 'waspy' after IBM finished training them.)
It was also a very crisp environment, everything neat, well organized,
'just so'. Univac was much more down-to-earth, friendly, easy-going;
it was more than merely the dress code (IBM white shirts, dark suits,
Univac colored shirts, sport jackets, slacks).
Eric Chomko
Thank God for ASCII rather than having to deal with either EBCDIC or
Fieldata.
Eric Chomko
> Anne & Lynn Wheeler wrote:
>
> > A Complete History Of Mainframe Computing
>
> Hardly complete, unless the world consists entirely of North America.
That but also, no reference to mini computers, micros, maxis and
workstations and how they ALL adversely effected the mainframe market
at some point or another after mainframes enjoyed not only domination
but were the only ones in the beginning.
Eric
Walter Bushell
wrote:
Not Texas? That's the only part I want to give back to Mexico. Would
improve both nations.
Lawrence Statton
> Add in those German Beer ads (was it?) in Mexico that showed
> California, arizona, new mexico as part of Mexico..
Absolut vodka, as part of their "In an Absolut World ... " series,
showed a 1845 (pre mexican-american war) map.
--L
CBFalconer
They used to be just that. Various wars changed things.
Rich Alderson
> A Complete History Of Mainframe Computing
> http://www.tomshardware.com/picturestory/508-mainframe-computer-history.html
As I noted earlier today on the ClassicCmp mailing list, this is almost
entirely an IBM list. With the execption of the VAX 11/780
(super)minicomputer, nothing from any non-IBM company past 1970 was shown,
and only a DEC minicomputer, the PDP-8, was shown in the 1960s, although
the PDP-10 (which competed with the System 360/50) would have been much
more appropriate.
Complete and utter horseshit, and should be ignored as same.
--
Rich Alderson "You get what anybody gets. You get a lifetime."
ne...@alderson.users.panix.com --Death, of the Endless
Eugene Miya
>> Hardly complete, unless the world consists entirely of North America.
>
>Any claim to a "complete" history of mainframe computing for something
>that is less than a 20-volume work is, of necessity, pretentious.
The guy was a computer operator starting about the IBM 370/3081.
Arguably, he's one of Roger Blake's generalities of a dumb young kid
learning stuff.
>Others have noted that some North American computer makers weren't
>mentioned either. On the other hand, the PDP-8 was a minicomputer, not
>a mainframe.
Not only that VAX-11/780s weren't mainframes either.
Has a CPU bias (storage/memory hierarchy tends to be a better measure
in some ways).
He starts out better with more factual information as he progresses to
his experience, he snubs Intel a bit, he uses flowery adjectives on the
things which he personal experience. The pages are kind of like a high
school or community college report project.
Complete? No. Turn the British (Nick) on him.
>It would indeed have been nice to see mention of the KDF-9 or the ICL
>1900. Or the Cray-1, or NCR mainframes... but it was a reasonable
>short survey of the subject.
Hmm, the cray-1 wasn't a mainframe. Nor the 6600.
Needs systems thinking.
--
Looking for an H-912 (container).
Charlie Gibbs
<36da85fc-136b-4f44...@j12g2000vbl.googlegroups.com>,
pne.c...@comcast.net (Eric Chomko) writes:
> Thank God for ASCII rather than having to deal with either EBCDIC or
> Fieldata.
From Ted Nelson's "Computer Lib":
"ASCII and ye shall receive." -- the computer industry
"ASCII not, what your machine can do for you." -- IBM
Eric Chomko
> In article
> <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>
> pne.cho...@comcast.net (Eric Chomko) writes:
> > Thank God for ASCII rather than having to deal with either EBCDIC or
> > Fieldata.
>
> From Ted Nelson's "Computer Lib":
>
> "ASCII and ye shall receive." -- the computer industry
>
> "ASCII not, what your machine can do for you." -- IBM
IBM realized their own error with EBCDIC and made ASCII the character
code set for the PC.
Scott Lurndal
>On Jul 2, 11:56=A0am, "Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
>> In article
>> <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>>
>> pne.cho...@comcast.net (Eric Chomko) writes:
>> > Thank God for ASCII rather than having to deal with either EBCDIC or
>> > Fieldata.
>>
>> From Ted Nelson's "Computer Lib":
>>
>>
>> "ASCII not, what your machine can do for you." =A0-- IBM
>
>IBM realized their own error with EBCDIC and made ASCII the character
>code set for the PC.
How, exactly, was EBCDIC an error, particularly since it was always 8-bit
while ASCII wasn't? One can represent all the same characters, and convertion
from upper case to lower case (or vise versa) is as simple as setting (or resetting)
a bit.
scott
Rostyslaw J. Lewyckyj
> On Jul 2, 11:56 am, "Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
>> In article
>> <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>>
>> pne.cho...@comcast.net (Eric Chomko) writes:
>>> Thank God for ASCII rather than having to deal with either EBCDIC or
>>> Fieldata.
>> From Ted Nelson's "Computer Lib":
>>
>> "ASCII and ye shall receive." -- the computer industry
>>
>> "ASCII not, what your machine can do for you." -- IBM
>
> IBM realized their own error with EBCDIC and made ASCII the character
> code set for the PC.
>
And why did it not get used?
Quadibloc
> Hmm, the cray-1 wasn't a mainframe. Nor the 6600.
They weren't minicomputers, or microcomputers. So they were
mainframes. What else is there? A supercomputer was simply the largest
kind of mainframe.
The 6600, at least, had auxiliary processors for I/O - basically, one
12-bit multithreaded computer. They were scientific mainframes instead
of commercial mainframes.
John Savard
Quadibloc
> Thank God for ASCII rather than having to deal with either EBCDIC or
> Fieldata.
Or CDC Display Code.
John Savard
Quadibloc
wrote:
> So what was bit 12 of the IBM360 PSW for?
> And why did it not get used?
Because it switched the 360 into using a funny form of 8-bit ASCII
that nobody else ever used. And IBM didn't make software that ran in
that mode anyways.
John Savard
Patrick Scheible
> On Jul 1, 6:43=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>
> > Hmm, the cray-1 wasn't a mainframe. =A0Nor the 6600.
>
> They weren't minicomputers, or microcomputers. So they were
> mainframes. What else is there? A supercomputer was simply the largest
> kind of mainframe.
I don't agree with that definition. Mainframes are meant to
economically solve data processing needs. In supercomputers,
price/performance is secondary because the problems are so big that
lesser computers can't solve them in a timely way.
-- Patrick
Walter Bushell
<092db050-9024-4732...@d4g2000yqa.googlegroups.com>,
Eric Chomko <pne.c...@comcast.net> wrote:
> On Jul 2, 11:56�am, "Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
> > In article
> > <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
> >
> > pne.cho...@comcast.net (Eric Chomko) writes:
> > > Thank God for ASCII rather than having to deal with either EBCDIC or
> > > Fieldata.
> >
> > From Ted Nelson's "Computer Lib":
> >
> > "ASCII and ye shall receive." �-- the computer industry
> >
> > "ASCII not, what your machine can do for you." �-- IBM
>
> IBM realized their own error with EBCDIC and made ASCII the character
> code set for the PC.
The PC group was a renegade group by IBM standards and was allowed to
flout company standards because they were so cute, and could hardly
challenge IBM mainlines.
ArarghMai...@not.at.arargh.com
wrote:
I thought all it did was change the sign value for packed decimal
instructions.
--
ArarghMail907 at [drop the 'http://www.' from ->] http://www.arargh.com
BCET Basic Compiler Page: http://www.arargh.com/basic/index.html
To reply by email, remove the extra stuff from the reply address.
![Tony Toews [MVP]'s profile photo](http://lh3.googleusercontent.com/a-/ALV-UjVFNFhAO8kxGtutFBSPYzx5Lbrrty7WRKPHNhjuWtTEHvsoQdk=s40-c)
Tony Toews [MVP]
Where is the dividing line between mainframe, mini and micro? Now with cell phones
having more computing power than mainframes from the 70s then adds the question.
What year or decade do you make the distinctions between mainframe, mini and micro?
Tony
--
Tony Toews, Microsoft Access MVP
Tony's Main MS Access pages - http://www.granite.ab.ca/accsmstr.htm
Tony's Microsoft Access Blog - http://msmvps.com/blogs/access/
Granite Fleet Manager http://www.granitefleet.com/
Quadibloc
> On Thu, 2 Jul 2009 18:30:11 -0700 (PDT), Quadibloc <jsav...@ecn.ab.ca>
> wrote:
> >On Jul 2, 5:08 pm, "Rostyslaw J. Lewyckyj" <urj...@bellsouth.net>
> >wrote:
>
> >> So what was bit 12 of the IBM360 PSW for?
> >> And why did it not get used?
>
> >Because it switched the 360 into using a funny form of 8-bit ASCII
> >that nobody else ever used. And IBM didn't make software that ran in
> >that mode anyways.
>
> I thought all it did was change the sign value for packed decimal
> instructions.
Well, it also changed the character codes that digits got translated
to when you unpacked decimals. And the resulting printable characters
for digits were in "ASCII-8" illustrated in a nice table at the back
of the Principles of Operation. (Turns around, reaches for his copy.)
Oh, that was USASCII-8.
Bits 7,6,5,4,3,2,1 were converted to 7,6,7,5,4,3,2,1 in this code so
that it would have the same space character as EBCDIC. As they nicely
explain in Appendix F.
John Savard
Anne & Lynn Wheeler
Patrick Scheible <k...@zipcon.net> writes:
> I don't agree with that definition. Mainframes are meant to
> economically solve data processing needs. In supercomputers,
> price/performance is secondary because the problems are so big that
> lesser computers can't solve them in a timely way.
modulo more recent migration to off-the-self (cots) parts ... old
email about cluster scaleup
http://www.garlic.com/~lynn/lhwemail.html#medusa
in ha/cmp scaleup
http://www.garlic.com/~lynn/subtopic.html#hacmp
then the scaleup activity got transferred and we were told that we
couldn't work on anything with more than four processors.
within weeks after the transfer ... some press releases:
http://www.garlic.com/~lynn/2001n.html#6000clusters1 ... scientific and technical *only*
http://www.garlic.com/~lynn/2001n.html#6000clusters2 ... caught by *surprise*
recent items in the trend:
News Story DARPA Wants a 19" Super-Efficient Supercomputer
http://news.slashdot.org/story/09/06/27/2118232/DARPA-Wants-a-19-Super-Efficient-Supercomputer
Supercomputers Lose Glamour, Price Tag
weekly.rup:http://www.pcworld.com/businesscenter/article/167504/supercomputers_lose_glamour_price_tag.html
reference to writing several papers in early 1985 about cluster
configuration "densely" packing as many microprocessors as possible in
racks (major problem was heat and cooling)
http://www.garlic.com/~lynn/2004m.html#17 mainframe and microprocessor
--
40+yrs virtualization experience (since Jan68), online at home since Mar1970
ArarghMai...@not.at.arargh.com
wrote:
>On Jul 2, 10:48�pm, ArarghMail907NOS...@NOT.AT.Arargh.com wrote:
Ok. But still only seems to affect packed decimal instructions.
Quadibloc
> Ok. But still only seems to affect packed decimal instructions.
Yes. There aren't any others which give semantic meaning to characters
to be affected.
John Savard
Brian Boutel
>
> Where is the dividing line between mainframe, mini and micro? Now with cell phones
> having more computing power than mainframes from the 70s then adds the question.
> What year or decade do you make the distinctions between mainframe, mini and micro?
>
Until the late 60s there were just "computers". The advent of the
"mini-computer" forced a distinction, which was that (generalising)
minis occupied a single cabinet/frame, while big computers spread over
several, connected by underfloor cabling. The separate cabinet that held
the cpu was called the "main frame", so a "mainframe" computer was
simply a computer that had one.
"Microcomputers" were those powered by a microprocessor, i.e. an early,
probably 8-bit, cpu on a single chip. (The lsi-11 cpu was on 4 chips, so
probably doesn't count as a micro.)
Later, people re-classified machines so that a mainframe was defined
according to its typical use in data processing, with lots of I/O grunt
and not so much cpu power, so excluding powerful timesharing machines
and supercomputers. That's a fair distinction, but a misuse of the term.
--brian
(who is about to head off for Europe to see his grandson and won't be
reading news for a while.)
--
Wellington, New Zealand
"I don't respond to Christopher Hitchens in public, on the general
principle that you should never mud-wrestle with a pig because you both
get filthy and the pig likes it." -- Tony Judt
Anne & Lynn Wheeler
Brian Boutel <fa...@fake.nz> writes:
> Until the late 60s there were just "computers". The advent of the
> "mini-computer" forced a distinction, which was that (generalising)
> minis occupied a single cabinet/frame, while big computers spread over
> several, connected by underfloor cabling. The separate cabinet that
> held the cpu was called the "main frame", so a "mainframe" computer
> was simply a computer that had one.
re:
http://www.garlic.com/~lynn/2009j.html#22 A Complete History Of Mainframe Computing
http://www.garlic.com/~lynn/2009j.html#37 A Complete History Of Mainframe Computing
in the discussion about 701 and 650 ... 701 was "defense calculator"
... and only 19 sold ... while 650 was something of a more sophisticated
tabulating equipment ... used in conjunction with other card tabulating
equipment (and sold 2000). one may conjecture that increasing
sophistication of 650 and follow-ons, "mainframe" might be used to
differentiate from the other card tabulating equipment that would also
be around.
recent 701 & 650 thread
http://www.garlic.com/~lynn/2009h.html#12 IBM Mainframe: 50 Years of Big Iron Innovation
Tony Toews [MVP]
>> Where is the dividing line between mainframe, mini and micro? Now with cell phones
>> having more computing power than mainframes from the 70s then adds the question.
>> What year or decade do you make the distinctions between mainframe, mini and micro?
>>
>
>Until the late 60s there were just "computers". The advent of the
>"mini-computer" forced a distinction, which was that (generalising)
>minis occupied a single cabinet/frame, while big computers spread over
>several, connected by underfloor cabling. The separate cabinet that held
>the cpu was called the "main frame", so a "mainframe" computer was
>simply a computer that had one.
And yet the IBM S/38 had the 3370 DASD in a separate box. The rack mount AS/400
could have a number of racks. So the dividing line is quite faint as they were both
certainly known as a mini computer.
>"Microcomputers" were those powered by a microprocessor, i.e. an early,
>probably 8-bit, cpu on a single chip. (The lsi-11 cpu was on 4 chips, so
>probably doesn't count as a micro.)
Reasonable enough.
Mensanator
> Tony Toews [MVP] wrote:
>
> > Where is the dividing line between mainframe, mini and micro? � �Now with cell phones
> > having more computing power than mainframes from the 70s then adds the question.
> > What year or decade do you make the distinctions between mainframe, mini and micro?
>
> Until the late 60s there were just "computers". The advent of the
> "mini-computer" forced a distinction, which was that (generalising)
> minis occupied a single cabinet/frame, while big computers spread over
> several, connected by underfloor cabling. The separate cabinet that held
> the cpu was called the "main frame", so a "mainframe" computer was
> simply a computer that had one.
>
> "Microcomputers" were those powered by a microprocessor, i.e. an early,
> probably 8-bit, cpu on a single chip.
The CPU may have been a single chip, but you're
ingnoring the myriad support chips like timers,
counters, serial I/O, parallel I/O, etc. I'm mostly
referring to 8080 and Z80 but how big a chunk do
they represent?
>(The lsi-11 cpu was on 4 chips, so
> probably doesn't count as a micro.)
Sure it does. Four 40-pin chips compared to
four cards, each a foot square is quite a
difference between a mini and a micro.
And, if I recall, one of those 4 chips was
an FPU, the LSI-11s I used only had 3 chips
as we didn't buy the FPU option.
There was a microprocessor I know of that
had everything in one 40-pin package (Signetics
2650) but as it had limitations (only 14 address
lines), I know only of its use as a controller.
hen again, there were the 4-bit bit-slice units
(AMD 2900) where 4 chips combined to make a
16-bit processor.
Eugene Miya
Quadibloc <jsa...@ecn.ab.ca> wrote:
>On Jul 1, 6:43=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>> Hmm, the cray-1 wasn't a mainframe. =A0Nor the 6600.
>
>They weren't minicomputers, or microcomputers. So they were
>mainframes. What else is there? A supercomputer was simply the largest
>kind of mainframe.
I'll get to this section above with Patrick's post.
>The 6600, at least, had auxiliary processors for I/O - basically, one
Sure, so did the Cray-1. They had polled I/O.
>12-bit multithreaded computer. They were scientific mainframes instead
>of commercial mainframes.
So as I pointed out to my JPL boss (one of the best I ever had: we Never
had group meetings) when I was running on Univac 1108s and IBM 360/75s and
a 3032: what (in a "scientific" machine) makes you think
these are any different from a "commercial" machine? What do you think
makes a machine a mainframe? Think "frame."
People have vague notions of the CPU, practically no one (I'd bet dmr
could) could name a Cray disk. Much less the nature of the wiring,
performance specs w/o resorting to web searches (my friend Garb could,
maybe Fair when he was at Apple).
Eugene Miya
Memory is/was a major component at the time. Economy only kind of
works into it to a degree.
The problem is that you guys are concentrating on the CPU. Cray did
systems. The problem with this fellow's web site was his lack of
understanding the 1604. Even Seymour's 1988 address, which I recently
relistened to, Cray mentioned the 1604 and paused and the audience just
didn't get it (which was something I could tell being there). To this
day most people could not thread the proper chain of firms Cray worked
for and what they do/did. Likely, there are special machines done which
we may never hear of. And a few we only peripherally may encounter.
John's problem is that he's mapping everything to a line (an axis).
He's had no experience with anything other than a von Neuamnn machine.
Eugene Miya
Close. Real close.
I forgot how much they (IBM) said it would cost to simply produce an empty box.
Patrick Scheible
> In article <w9zocs2...@zipcon.net>,
> Patrick Scheible <k...@zipcon.net> wrote:
> >Quadibloc <jsa...@ecn.ab.ca> writes:
> >> On Jul 1, 6:43=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
> >> > Hmm, the cray-1 wasn't a mainframe. =A0Nor the 6600.
> >>
> >> They weren't minicomputers, or microcomputers. So they were
> >> mainframes. What else is there? A supercomputer was simply the largest
> >> kind of mainframe.
> >
> >I don't agree with that definition. Mainframes are meant to
> >economically solve data processing needs. In supercomputers,
> >price/performance is secondary because the problems are so big that
> >lesser computers can't solve them in a timely way.
>
> Memory is/was a major component at the time. Economy only kind of
> works into it to a degree.
Sure memory is important. However, I was trying to speak more
generally about what made a computer 'super', regardless of what era
it was.
> The problem is that you guys are concentrating on the CPU.
Um. I deliberately avoided the CPU (or CPUs) in how I characterized
supercomputers.
-- Patrick
Joe Pfeiffer
There were no "mainframes" until there were minicomputers to contrast
them with. When that happened, the distinguishing characteristics were
a separate IO bus and separate IO processors. So, if you're willing to
accept this definition, a 6600 wasn't just a mainframe, its offload of the
OS made it a super-mainframe (appropriately enough)
Peter Flass
>
> John's problem is that he's mapping everything to a line (an axis).
> He's had no experience with anything other than a von Neuamnn machine.
>
I doubt many of us do. Have any existed since the 50's?
Charles Richmond
>
> [snip...] [snip...] [snip...]
>
> There were no "mainframes" until there were minicomputers to contrast
> them with. When that happened, the distinguishing characteristics were
> a separate IO bus and separate IO processors. So, if you're willing to
> accept this definition, a 6600 wasn't just a mainframe, its offload of the
> OS made it a super-mainframe (appropriately enough)
It's kind of like: Michael Douglas used to be Kirk Douglas' son. Now
Kirk Douglas is Michael Douglas' father. The emphasis is different.
Back in the 60's, there were radios and transistor radios. Now there are
radios and tube radios. The assumption today is that radios are made
with transistors (or integrated circuits).
--
+----------------------------------------------------------------+
| Charles and Francis Richmond richmond at plano dot net |
+----------------------------------------------------------------+
Scott Lurndal
There was a considerable amount of research into data-flow
architectures in the 70's and 80's. I seem to recall hearing
of at least one machine being built.
I've done at least one software product that was internally a
data-flow engine, albeit built on a von Neumann architecture.
scott
Anne & Lynn Wheeler
bits & pieces from ha/cmp project marketing study we had done in fall of
89
Estimated 1988-1992 WW Installed Base
Category 1988 1992 CGR
Supercomputers 350 1000 22%
Mainframes 6000 12000 6%
minisupers 2200 11000 41%
superminis 300,000 900,000 20%
workstations 370,000 2,750,000 52%
Workstation Unit Shipments
Worldwide 1986-1993
(000s)
1986 50
1987 100
1988 200
1989 300
1990 450
1991 650
1992 950
1993 1200
misc. past posts mentioning ha/cmp
http://www.garlic.com/~lynn/subtopic.html#hacmp
then "supercomputers" turned into large collections of "workstation"
technology.
...
and misc. DEC "mainframe" news tidbits from Oct89:
DEC is planning a big coming-out party on Tuesday
- first line of mainframe computers ... VAX 9000
- IBM is expected to announce its own mainframe
. "They're deliberately trying to steal our thunder ....
maybe we should take it as a compliment" -DEC executive
- Digital has spent almost $1 billion developing new technology
. promises a new approach ... a decentralized network
. air-cooled mainframe
. It's been a long time coming
- A full family will be announced
. 4 models aimed squarely at IBM's commercial customers
. volume shipments won't begin until next year
- Model 210 will be the first available
. Appeals to technical and scientific buyers
. Optional "Supercharger" vector processor
- A spoiler, nimble Tandem, jumped into the fray earlier this week
. IBM seems more worried about DEC
Mainframe market .. $40 billion
- nearly every big company needs one
- IBM has dominated ... that doesn't scare DEC (world's 2nd largest mfger)
- Challengers have a big price advantage
. DEC: $1.24-4.4 million, and up (half the price of IBM)
. Tandem: Pricing is just as aggressive
- Competition hits IBM at a difficult time
. current mainframe line is starting to show its age
. New 3090's will boost performance only 8-10%
. New generation not expected from IBM until 1991
- No one expects competitors to deliver a knock-out punch to IBM
. IBM has a near monopoly on mainframes (est. 70%)
. IBM is 5-times the size of DEC; 40-times the size of Tandem
. IBM customers are the largest corporations in the world
- Kenneth Olsen
. "We're not going to walk in and replace a company's corporate
accounting system if it's already running on an IBM mainframe"
- DEC will target growing market segments
. online transaction processing (Tandem specializes in this market)
. customers who have both IBM and DEC equipment
Roland Hutchinson
> Joe Pfeiffer wrote:
>>
>> [snip...] [snip...] [snip...]
>>
>> There were no "mainframes" until there were minicomputers to contrast
>> them with. When that happened, the distinguishing characteristics were
>> a separate IO bus and separate IO processors. So, if you're willing to
>> accept this definition, a 6600 wasn't just a mainframe, its offload of
>> the OS made it a super-mainframe (appropriately enough)
>
> It's kind of like: Michael Douglas used to be Kirk Douglas' son. Now
> Kirk Douglas is Michael Douglas' father. The emphasis is different.
This, they stole from Abraham "Fr�her war ich der Sohn meines Vaters, jetzt
bin ich der Vater meines Sohnes" Mendelssohn.
--
Roland Hutchinson
He calls himself "the Garden State's leading violist da gamba,"
... comparable to being ruler of an exceptionally small duchy.
--Newark (NJ) Star Ledger ( http://tinyurl.com/RolandIsNJ )
Patrick Scheible
> bits & pieces from ha/cmp project marketing study we had done in fall of
> 89
>
> Estimated 1988-1992 WW Installed Base
>
> Category 1988 1992 CGR
>
> Supercomputers 350 1000 22%
> Mainframes 6000 12000 6%
> minisupers 2200 11000 41%
> superminis 300,000 900,000 20%
> workstations 370,000 2,750,000 52%
What's the difference between a minisuper and a supermini?
-- Patrick
Lawrence Statton
Speak for yourself ... Harvard Architecture DSPs are alive and well in
the 2000s
--L
Anne & Lynn Wheeler
> What's the difference between a minisuper and a supermini?
re:
http://www.garlic.com/~lynn/2009j.html#52 A Complete History Of Mainframe Computing
there is some implication that superminicomputers might have
(programming/software) compatibility with minicomputers (DEC) and
minisupercomputers might have compatibility with supercomputers (cray or
others)
this old post has extract from Jan88 report/study
http://www.garlic.com/~lynn/2001b.html#55 Why SMP at all anymore?
which seems to slip back & forth between minisupers and superminis
Alliant 171
Celerity just shipping
Convex 200
ELXSI 80
FPS 365
Gould 6
Multiflow 5
Scientific 25
Computing
Supertek not shipping yet
... snip ...
which lists approx. 850 minisupers (for US at some pt before Jan88
report) ... while ha/cmp fall 1989 marketing study lists 2200 (world
wide in 1988) ... aka part of ha/cmp was focused on HA - high
availability ... ha/cmp also had cmp cluster scaleup focus.
2200 for mini (small) supercomputers is significantly fewer than the
300,000 for super (large) minicomputers.
Peter Flass
I guess I don't associate DSPs with computers.
Rich Alderson
> hen again, there were the 4-bit bit-slice units
> (AMD 2900) where 4 chips combined to make a
> 16-bit processor.
And 9 of them combine to make a 36-bit processor (cf. the KS-10 processor in
a DECSYSTEM-2020).
--
Rich Alderson "You get what anybody gets. You get a lifetime."
ne...@alderson.users.panix.com --Death, of the Endless
Rich Alderson
> Until the late 60s there were just "computers". The advent of the
> "mini-computer" forced a distinction
The advent of the minicomputer was the *early* 1960s...
Eugene Miya
Patrick Scheible <k...@zipcon.net> wrote:
>eug...@cse.ucsc.edu (Eugene Miya) writes:
>> In article <w9zocs2...@zipcon.net>,
>> Patrick Scheible <k...@zipcon.net> wrote:
>> >Quadibloc <jsa...@ecn.ab.ca> writes:
>> >> On Jul 1, 6:43=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>> >> > Hmm, the cray-1 wasn't a mainframe. =A0Nor the 6600.
>> >>
>> >> They weren't minicomputers, or microcomputers. So they were
>> >> mainframes. What else is there? A supercomputer was simply the largest
>> >> kind of mainframe.
>> >
>> >I don't agree with that definition. Mainframes are meant to
>> >economically solve data processing needs. In supercomputers,
>> >price/performance is secondary because the problems are so big that
>> >lesser computers can't solve them in a timely way.
>>
>> Memory is/was a major component at the time. Economy only kind of
>> works into it to a degree.
>
>Sure memory is important. However, I was trying to speak more
>generally about what made a computer 'super', regardless of what era
>it was.
As my honorable ancestors learned, we Americans are very slippery folk,
linguistically.
The prefix "super" from the comic character from the 30s planted itself
in Ed Teller's brain in WWII for his thoughts of thermonuclear fusion.
It was an idea (Bethe's) whose time had come, but Oppenheimer had enough
to think about just getting fission to work with Teller saying "super",
"super" in his ear all the time. And in fact a contemporary big
computer was needed at the various points for Monte Carlo and other work.
LASL (and then later LLL) probably went about a decade before prepending
super- in front of computer, for the 6600. Despite what one can read on
Wikipedia. And that's apparently where the MFLOPS metric also stuck in
Congressional testimony to ask for money for the machine. Cray, for
instance never used the prefix, and he refused to use it. It started
with a bunch of guys led by one Sid Fernbach (with Teller behind him of
course). And Sid wasn't satisifed, so he gave the world "Classes" of
supercomputer. Which largely ended at Class 6 (or VI) with
a Cray 1 or X-MP or maybe even a Cyber 205 (4 pipes was best and
you had to have certain kinds of codes; and we had the only
4 pipe 205 for political reasons: did that mean all the 2-pipe and 1-pipe
machines weren't supers? the MN Congressional delegation won't have that).
>> The problem is that you guys are concentrating on the CPU.
>
>Um. I deliberately avoided the CPU (or CPUs) in how I characterized
>supercomputers.
What basically made a super "super" was some emotional act of
desparation which would let Congress give you money to buy the thing.
Sid was a character. And he wrote a book. Not that I would agree with
all portions of that book. He died on a day when a bunch of us were
waiting for him at Tadish Grill in SF.
Basically think of nasty ugly uncompromising codes (application programs
which might take machine years to execute) and those are programs for
these kinds fo machines. Up to a point.
Eugene Miya
Peter Flass <Peter...@Yahoo.com> wrote:
>Eugene Miya wrote:
>> John's problem is that he's mapping everything to a line (an axis).
>
>I doubt many of us do. Have any existed since the 50's?
Many.
I was thinking about the Systron-Donner analog machine last evening
(chaos theory). CHM collection problem.
We had an optical correlator for SAR (have to write that up still) at JPL.
Another history problem.
Associative processors on board the AWACS, etc.
Eugene Miya
Joe Pfeiffer <pfei...@cs.nmsu.edu> wrote:
>Patrick Scheible <k...@zipcon.net> writes:
>> eug...@cse.ucsc.edu (Eugene Miya) writes:
>>> In article <w9zocs2...@zipcon.net>,
>>> Patrick Scheible <k...@zipcon.net> wrote:
>>> >Quadibloc <jsa...@ecn.ab.ca> writes:
>>> >> On Jul 1, 6:43=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>>> >> > Hmm, the cray-1 wasn't a mainframe. =A0Nor the 6600.
>>> >> They weren't minicomputers, or microcomputers. So they were
>>> >> mainframes. What else is there? A supercomputer was simply the largest
>>> >> kind of mainframe.
>>> >I don't agree with that definition. Mainframes are meant to
>> generally about what made a computer 'super', regardless of what era
>> it was.
>>
>>> The problem is that you guys are concentrating on the CPU.
>>
>> Um. I deliberately avoided the CPU (or CPUs) in how I characterized
>> supercomputers.
>
>There were no "mainframes" until there were minicomputers to contrast
>them with. When that happened, the distinguishing characteristics were
>a separate IO bus and separate IO processors. So, if you're willing to
>accept this definition, a 6600 wasn't just a mainframe, its offload of the
>OS made it a super-mainframe (appropriately enough)
I'd go with that.
Minis had smaller address spaces and still did useful, if not all, work.
Minis were more important after a certain period than mainframes anyways.
And then later micros.
The LINC which begat the PDP-1 were all done ahead of the old Soviet
Union which realized their utility way too late.
Eugene Miya
Scott Lurndal <sl...@pacbell.net> wrote:
>Peter Flass <Peter...@Yahoo.com> writes:
>>Eugene Miya wrote:
>>> John's problem is that he's mapping everything to a line (an axis).
>>> He's had no experience with anything other than a von Neuamnn machine.
>>
>>I doubt many of us do. Have any existed since the 50's?
>
>There was a considerable amount of research into data-flow
>architectures in the 70's and 80's. I seem to recall hearing
>of at least one machine being built.
Oh there were at least a dozen.
I shared an office with Jack Dennis for a year. We were the only tea
drinkers in my group. We worked with the LLNL and DEC and CSU FC
dataflow groups on the Manchester machine. The French had 1-2. Several
Japanese data flow machines were made (messes of cabling).
>I've done at least one software product that was internally a
>data-flow engine, albeit built on a von Neumann architecture.
DF means different things to different people. Search for single
assignment. vN engines were common building blocks.
Eugene Miya
Jobs had the foresight to place one in the NeXT.
These kinds of peripherals are the kinds of things small real networked
labs are needing. Not rooms full of PCs.
Eugene Miya
Anne & Lynn Wheeler <ly...@garlic.com> wrote:
>Estimated 1988-1992 WW Installed Base
>
>Category 1988 1992 CGR
>
>Supercomputers 350 1000 22%
>Mainframes 6000 12000 6%
>minisupers 2200 11000 41%
>superminis 300,000 900,000 20%
>workstations 370,000 2,750,000 52%
>
>
>http://www.garlic.com/~lynn/subtopic.html#hacmp
>
>then "supercomputers" turned into large collections of "workstation"
>technology.
>
>...
>
>and misc. DEC "mainframe" news tidbits from Oct89:
>
>DEC is planning a big coming-out party on Tuesday
>- first line of mainframe computers ... VAX 9000
First? Compared to the claimed 10s and 20s.
Point in my favor.
ML-02 considered the 9000 a Cray-1 equivalent.
I presume non-disclosure is over.
>- IBM is expected to announce its own mainframe
> . "They're deliberately trying to steal our thunder ....
> maybe we should take it as a compliment" -DEC executive
Oh.
>- Digital has spent almost $1 billion developing new technology
> . promises a new approach ... a decentralized network
> . air-cooled mainframe
> . It's been a long time coming
They only had the 2 CPU 8800 at the time as their previous high end.....
>- A full family will be announced
> . 4 models aimed squarely at IBM's commercial customers
> . volume shipments won't begin until next year
>- Model 210 will be the first available
> . Appeals to technical and scientific buyers
> . Optional "Supercharger" vector processor
>- A spoiler, nimble Tandem, jumped into the fray earlier this week
> . IBM seems more worried about DEC
>
>Mainframe market .. $40 billion
>- nearly every big company needs one
>- IBM has dominated ... that doesn't scare DEC (world's 2nd largest mfger)
>- Challengers have a big price advantage
> . DEC: $1.24-4.4 million, and up (half the price of IBM)
> . Tandem: Pricing is just as aggressive
>- Competition hits IBM at a difficult time
> . current mainframe line is starting to show its age
> . New 3090's will boost performance only 8-10%
> . New generation not expected from IBM until 1991
>- No one expects competitors to deliver a knock-out punch to IBM
> . IBM has a near monopoly on mainframes (est. 70%)
> . IBM is 5-times the size of DEC; 40-times the size of Tandem
> . IBM customers are the largest corporations in the world
Yep.
Quite true.
>- Kenneth Olsen
> . "We're not going to walk in and replace a company's corporate
> accounting system if it's already running on an IBM mainframe"
Yep.
Quite true.
>- DEC will target growing market segments
> . online transaction processing (Tandem specializes in this market)
That didn't work.
> . customers who have both IBM and DEC equipment
--
Looking for an H-912 (container).
Al Kossow
> CHM collection problem.
That we are out of storage space?
Charles Richmond
> Mensanator <mensa...@aol.com> writes:
>
>
>> hen again, there were the 4-bit bit-slice units
>> (AMD 2900) where 4 chips combined to make a
>> 16-bit processor.
>
> And 9 of them combine to make a 36-bit processor (cf. the KS-10 processor in
> a DECSYSTEM-2020).
>
ISTM that the Harris 800 computer used AMD 2900's on their CPU board...
Joe Pfeiffer
Not general purpose, of course, but they certainly meet any reasonable
definition of "computer".
Eric Chomko
> Eric Chomko <pne.cho...@comcast.net> writes:
> >On Jul 2, 11:56=A0am, "Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
> >> In article
> >> <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>
> >> pne.cho...@comcast.net (Eric Chomko) writes:
> >> > Thank God for ASCII rather than having to deal with either EBCDIC or
> >> > Fieldata.
>
> >> From Ted Nelson's "Computer Lib":
>
>
> >> "ASCII not, what your machine can do for you." =A0-- IBM
>
> >IBM realized their own error with EBCDIC and made ASCII the character
> >code set for the PC.
>
> while ASCII wasn't? One can represent all the same characters, and convertion
> from upper case to lower case (or vise versa) is as simple as setting (or resetting)
> a bit.
>
Compare the price of EBCDIC terminals to their ASCII counterparts. Its
the same thing when comparing PL/I compilers to C compilers. IBM owned
EBCDIC. The world owns ASCII.
EBCDIC wasn't inherently bad other than the fact that IBM owned it.
ASCII eventually had 8 bits with extended ASCII.
Eric Chomko
> Eric Chomko <pne.cho...@comcast.net> wrote:
>
>
>
>
>
> > On Jul 2, 11:56 am, "Charlie Gibbs" <cgi...@kltpzyxm.invalid> wrote:
> > > In article
> > > <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>
> > > pne.cho...@comcast.net (Eric Chomko) writes:
> > > > Thank God for ASCII rather than having to deal with either EBCDIC or
> > > > Fieldata.
>
> > > From Ted Nelson's "Computer Lib":
>
>
> > > "ASCII not, what your machine can do for you." -- IBM
>
> > IBM realized their own error with EBCDIC and made ASCII the character
> > code set for the PC.
>
> flout company standards because they were so cute, and could hardly
> challenge IBM mainlines.
The Boca Project was a big secret operation for IBM. And yes, the PC
was considered something not quite in the same league as the
mainframe.
Eric Chomko
> Anne & Lynn Wheeler <l...@garlic.com> wrote:
>
> Where is the dividing line between mainframe, mini and micro? Now with cell phones
> having more computing power than mainframes from the 70s then adds the question.
> What year or decade do you make the distinctions between mainframe, mini and micro?
>
Well mainframes are bigger than maxis and maxis are bigger than minis.
Minis are bigger than workstations but aren't as powerful.
Workstations are bigger than micros and used to be more powerful but
now workstations don't exist anymore because micros made them
obsolete. So today there are no maxis, minis or workstations and only
micros and mainframes. And the main difference between a micro and a
mainframe is scalability.
There that should keep this thread going for awhile. :)
Eric
Eric Chomko
wrote:
> Brian Boutel <f...@fake.nz> writes:
> > Until the late 60s there were just "computers". The advent of the
> > "mini-computer" forced a distinction
>
> The advent of the minicomputer was the *early* 1960s...
More to the original poster's point, Webster states "minicomputer" was
first used in 1967. http://www.merriam-webster.com/dictionary/minicomputer
Eric
>
> --
> Rich Alderson "You get what anybody gets. You get a lifetime."
> n...@alderson.users.panix.com --Death, of the Endless
Esra Sdrawkcab
wrote:
Well it did (does) have the alphabet scattered about, making a "is this a
letter" test a bit tricky.
--
Siggy played guitar
Joe Pfeiffer
I would regard a character set in which neither the lower case
nor upper case letters are contiguous to be pretty desperately flawed.
Quadibloc
> I would regard a character set in which neither the lower case
> nor upper case letters are contiguous to be pretty desperately flawed.
Well, it was based on the pre-existing code used on punched cards.
Of course it would be preferable for certain purposes for the letters
to be contiguous, as they are in ASCII. It, too, however, is
imperfect. A does not come after 9.
John Savard
Eric Chomko
Okay, EBCDIC is inherently flawed. I thought about range checks on
printable charaters (as well as letters) and loops for processing
characters, and agree.
Scott Lurndal
>On Jul 2, 6:16=A0pm, sc...@slp53.sl.home (Scott Lurndal) wrote:
>> Eric Chomko <pne.cho...@comcast.net> writes:
>> >> In article
>> >> <36da85fc-136b-4f44-ade4-71bbee795...@j12g2000vbl.googlegroups.com>,
>>
>> >> pne.cho...@comcast.net (Eric Chomko) writes:
>> >> > Thank God for ASCII rather than having to deal with either EBCDIC or
>> >> > Fieldata.
>>
>> >> From Ted Nelson's "Computer Lib":
>>
>>
>> >> "ASCII not, what your machine can do for you." =3DA0-- IBM
>>
>> >IBM realized their own error with EBCDIC and made ASCII the character
>> >code set for the PC.
>>
>> How, exactly, was EBCDIC an error, particularly since it was always 8-bit
>vertion
>> from upper case to lower case (or vise versa) is as simple as setting (or=
> resetting)
>> a bit.
>>
>
>Compare the price of EBCDIC terminals to their ASCII counterparts. Its
>the same thing when comparing PL/I compilers to C compilers. IBM owned
>EBCDIC. The world owns ASCII.
Neither of these analogies support your contention that EBCDIC was
a worse character encoding than ASCII.
>
>EBCDIC wasn't inherently bad other than the fact that IBM owned it.
>ASCII eventually had 8 bits with extended ASCII.
IBM didn't _own_ EBCDIC. It was also used by other Mainframe lines.
scott
Rich Alderson
> In article <m34oto8...@garlic.com>,
> Anne & Lynn Wheeler <ly...@garlic.com> wrote:
>> and misc. DEC "mainframe" news tidbits from Oct89:
>> DEC is planning a big coming-out party on Tuesday
>> - first line of mainframe computers ... VAX 9000
> First? Compared to the claimed 10s and 20s.
> Point in my favor.
Not much of one. The Digital marketing folks were raked over the coals by
the 36-bit user crowd for that one, at DECUS, immediately following the 9000
announcement.
In point of fact, the 25th Anniversary T-shirt (from the same DECUS) listed
"DEC's First Mainframes" on the back. I know. I designed the damn thing.
(The front was done by another member of the Working Group which I chaired.)
I wore mine to the VMS Magic Session, and got a CCTV cameraman to put a shot
of the T-shirt up on the projection screens around the room. Interesting
crowd reaction...
--
Rich Alderson "You get what anybody gets. You get a lifetime."
Rich Alderson
> The LINC which begat the PDP-1 were all done ahead of the old Soviet
> Union which realized their utility way too late.
The LINC was not started until more than a year after the introduction of the
PDP-1. You're confusing it with the TX-0 and TX-2.
Peter Flass
Never caused me any problems, and still doesn't. Why do you need them
contiguous?
if verify(char,'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ')>0
then put('not an alphabetic character');
Peter Flass
More importantly, 0 does not come after Z (or z).
Quadibloc
Oh?
I'm not sure what purpose would be served by having the
sequence ...XYZ0123... in the code, but having ...789ABC... would mean
that conversion to hexadecimal would be even simpler and more natural
than conversion to decimal.
True, the traditional collation order in punched card systems had the
digits coming after the letters, but if one wants to preserve that,
one might as well stick with EBCDIC as is.
John Savard
Quadibloc
> Joe Pfeiffer wrote:
It requires a table, and if the table is not pre-formed, it would take
many more cycles.
John Savard
Eugene Miya
Naw, likely finding one.
Time would be better spent looking for other things. It depends how
clueless or clueful we want to be.
One can assemble a list, get a truck or whatever container, but if one
isn't aware about picking something up like say a PARSYTEC, then "That's tough."
Eugene Miya
Rich Alderson <ne...@alderson.users.panix.com> wrote:
>eug...@cse.ucsc.edu (Eugene Miya) writes:
>
>> In article <m34oto8...@garlic.com>,
>> Anne & Lynn Wheeler <ly...@garlic.com> wrote:
>>> and misc. DEC "mainframe" news tidbits from Oct89:
>>> DEC is planning a big coming-out party on Tuesday
>>> - first line of mainframe computers ... VAX 9000
>
>> First? Compared to the claimed 10s and 20s.
>> Point in my favor.
>
>Not much of one. The Digital marketing folks were raked over the coals by
>the 36-bit user crowd for that one, at DECUS, immediately following the 9000
>announcement.
The mistake the 9000 folk made was reading Russell's CACM paper when
they should have read the IEEE Computer X-MP article.
All they could do was apologize and show us an cardboard mock up.
No skin off my nose. I got to visit the Mill Pond. And Alliant, and
Encore (Multiflow had just folded). And various MIT labs.
>In point of fact, the 25th Anniversary T-shirt (from the same DECUS) listed
>"DEC's First Mainframes" on the back. I know. I designed the damn thing.
>(The front was done by another member of the Working Group which I chaired.)
>
>I wore mine to the VMS Magic Session, and got a CCTV cameraman to put a shot
>of the T-shirt up on the projection screens around the room. Interesting
>crowd reaction...
It's IBM's attitude to you guys. Get Lynn to change his text.
Eugene Miya
Rich Alderson <ne...@alderson.users.panix.com> wrote:
>eug...@cse.ucsc.edu (Eugene Miya) writes:
>> The LINC which begat the PDP-1 were all done ahead of the old Soviet
>> Union which realized their utility way too late.
>
>The LINC was not started until more than a year after the introduction of the
>PDP-1. You're confusing it with the TX-0 and TX-2.
Oops order.
Taylor really credits the LINC with helping DEC. I figure more the PDP-1.
But the windows of importance and inertia in markets aren't sharp
divisions. That guys in CA were assembling the first ICs which would
make Intel micros.
I never used either. I did attend the Workstation history meeting and
have dinner with Taylor a couple of times. And also taking to the old
officemate about his use of the PDP-1.
Eugene Miya
>> What's the difference between a minisuper and a supermini?
A super mini like the DEC VAX-11/780 or the Perkin-Elmer 8/32 (later
named Concurrent I think) were address scaled up minis like 16-bits,
plus or minus a few (20, 22, 24, etc.) to 32-bit.
>re:
>http://www.garlic.com/~lynn/2009j.html#52 A Complete History Of Mainframe Computing
>
>there is some implication that superminicomputers might have
>(programming/software) compatibility with minicomputers (DEC) and
Yeah compatbility mode.
>minisupercomputers might have compatibility with supercomputers (cray or
>others)
There were 2 classes of machines. Until the Cray X-MP, Crays were never
compatible. Come with software? Say what? You rolled your own OS.
You want speed, you bought what he offered.
Forrest Baskett wrote "A Unix-like" message passing operating system
named DEIMOS/DEMOS. Reports were that it was slow, like 10% performance
under existing OSes. Unix got a bad, and still has a bad reputation in
some quarters (I was called a Mother fucker as recently as oh maybe 3-4
years ago at Salishan by a somewhat noted astrophysicist ex-weaponeer,
but I was merely vanguard for what was inevitable). The reality is that
it takes 10 years to debug and tune OSes.
So what happened was that Seymour goes on to design the Cray 2. Never
compatible. But Cray software: operating systems and compilers are
starting to get long in the tooth. They are written in CAL (assembly
language). They need OSes and compilers in HLLs to stay ahead of 3
Japanese firms who studied at the UIUC and did their home work, ahead of
CDC (who didn't), and IBM (who bought their expertise and recently died).
So into the looming chasm come 3 firms: Scientific Computer Systems
(SCS-40 and SCS-30), American Supercomputer (Mike Flynn, really nice
guy), Supertek (Mike Fung, these begat Y-MP/ELs ultimately), and maybe
one other firm I forget its name.
They were 1/10 to 1/4 the speed of a Cray-1 or X-MP (CMOS-likely) and
object code compatible. COS/CAL and CFT (Fortran) were made public domain,
and the physics community (LANL/LLL, UIUC, SDSC (got a 40)) were going to run
their CTSS (Cray not Cambridge or Compatible). All these "minisupers"
were call by Datamation:
Crayettes
These were end application users getting ready to scramble and abandon ship.
Surprise.
Oh I wish I had been in Seattle when that announcement took place.
You could tell that people had really poor understanding of portability
and programming.
Convex even talked about writing a vector COBOL compiler which might
have scared IBM (they would never show sweat).
One has to be aware of things like Thoth, MDL, XXX (one of the Meade
software systems, I have to try to refind) and places like
NCAR (where Gombosi sent his last Cray days).
>this old post has extract from Jan88 report/study
>http://www.garlic.com/~lynn/2001b.html#55 Why SMP at all anymore?
>
>which seems to slip back & forth between minisupers and superminis
>
>Alliant 171
>Celerity just shipping
>Convex 200
>ELXSI 80
>FPS 365
That number is likely unclear because they had
unclear product lines.
>Gould 6
>Multiflow 5
>Scientific 25
> Computing
>Supertek not shipping yet
Early.
This leaves off lots of firms.
>which lists approx. 850 minisupers (for US at some pt before Jan88
>report) ... while ha/cmp fall 1989 marketing study lists 2200 (world
>wide in 1988) ... aka part of ha/cmp was focused on HA - high
>availability ... ha/cmp also had cmp cluster scaleup focus.
>
>2200 for mini (small) supercomputers is significantly fewer than the
>300,000 for super (large) minicomputers.
You guys rely on language poorly.
We all do.
Rostyslaw J. Lewyckyj
reason for the EBCDIC encoding was that it is an extension to 8 bits
from the 6 bits BCD encoding used in the previous generation of IBM
machines. The customers had a huge volume of BCD data on tape and
there apparently isn't an easy way read/convert from BCD to ASCII,
(something to do with either no unique 1-1 correspondence or
collation) . So the customer base objected to going to ASCII, which
resulted in the 4 sticks of 64 characters extension from BCD
i.e. EBCDIC encoding.
Consider the problems of handling 140x and possibly 707x and 709x
data tapes in emulation and interchange, especially during transition.
Bit 12 in the PSW was provided in a possibility of a later transition
to ASCII, which didn't materialize.
--
Rostyk
Rostyslaw J. Lewyckyj
Huh? In the US english EBCDIC encoding tables, code pages 037 & 500,
ISO/IEC 8859 equivalent - ISO/IEC 8859-1:1998
and others as well, Latin z is X'A9', Z in X'E9', and digit 0 is X'F0'
Since other language alphabets are all coded in the lower 128 i.e.
X'00' - X'7F', 0 = X'F0' will come after them.
------------
Oh I see you are criticizing ASCII. Never mind. :)
--
Rostyk
Chris Burrows
"Peter Flass" <Peter...@Yahoo.com> wrote in message
news:h33b2a$8b4$1...@news.eternal-september.org...
>
> Never caused me any problems, and still doesn't. Why do you need them
> contiguous?
>
> if verify(char,'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ')>0
> then put('not an alphabetic character');
Apart from the maintainability question (checking that the 52 characters are
correct takes longer than just checking two - or do you just assume the
original coder doesn't make typing errors?) this is much less efficient (in
both its usage of memory and CPU time) than is possible on an ASCII system:
IF ('A' < CAP(char)) OR (CAP(char) >'Z') THEN Out.String("not an alphabetic
character") END;
Maybe not a problem on an IBM mainframe, but for early PCs and even today in
embedded systems software it is a significant issue.
--
Chris Burrows
CFB Software
Armaide v2.0: ARM Oberon-07 Development System
http://www.cfbsoftware.com/armaide
Charles Richmond
Are you saying that EBCDIC is a "four letter word" ??? :-)
Joe Pfeiffer
Sure, you can work around it with a check for explicit characters. But
that test is so much simpler written as
if (((char >= 'A') && (char <= 'Z')) ||
((char >= 'a') && (char <= 'z')))
printf("alpha!");
I notice you assumed a verify() function, and didn't implement it.
Also, of course, I regard the very first principle of programming to be
"avoid circumstances that make bugs even easier" -- possibly a corollary
to the Act of Contrition, which states "I firmly resolve, with the help
of Thy grace, to sin no more and to avoid the near occasions of sin."
(yes, I am a Catholic).
Patrick Scheible
> So into the looming chasm come 3 firms: Scientific Computer Systems
> (SCS-40 and SCS-30), American Supercomputer (Mike Flynn, really nice
> guy), Supertek (Mike Fung, these begat Y-MP/ELs ultimately), and maybe
> one other firm I forget its name.
> They were 1/10 to 1/4 the speed of a Cray-1 or X-MP (CMOS-likely) and
> object code compatible. COS/CAL and CFT (Fortran) were made public domain,
> and the physics community (LANL/LLL, UIUC, SDSC (got a 40)) were going to run
> their CTSS (Cray not Cambridge or Compatible). All these "minisupers"
> were call by Datamation:
> Crayettes
Thank you for posting this history. I didn't know much about the
supers and minisupers side of the industry.
-- Patrick
Quadibloc
> IBM didn't _own_ EBCDIC. It was also used by other Mainframe lines.
IBM didn't patent or copyright EBCDIC. EBCDIC, though, was IBM's
creation; it wasn't devised by a standards committee somewhere and
subsequently adopted by IBM.
Its 6-bit predecessor, BCDIC, was also used by other computer makers,
while other codes based on the punched-card code were used by other
manufacturers. For example, there was one code which moved the letters
and digits downwards corresponding to excess-3 notation for decimal
digits.
John Savard
Peter Flass
It's the one true collating sequence;-)
Actually, I don't thing a character set matters much these days. The
problem I see with EBCDIC as a "standard" is that there are so many of them.
At work I got fed up with all the different codepages for different
applications and standardized everything on codepage 1047 (Latin-1 Open
Systems), which has a one-to-one completely reversible translation to ASCII.
Peter Flass
I certainly wouldn't check for alphabetic in any other way. Bit
twiddling or range checking is fine, but how do you handle accented
characters, which are also alphabetic? Using a table or, better yet,
internationalization functions to determine alphabetic is the only safe
way these days.
Peter Flass
It's a builtin in PL/I. As I said before, how do you handle accented
ASCII alphabetics? IIRC, they;re not contiguous with anything.
>
> Also, of course, I regard the very first principle of programming to be
> "avoid circumstances that make bugs even easier" -- possibly a corollary
> to the Act of Contrition, which states "I firmly resolve, with the help
> of Thy grace, to sin no more and to avoid the near occasions of sin."
> (yes, I am a Catholic).
Naturally, I'd code the alphabetic table once and include it when needed.
The other choice is to provide a translate table with bit values for
each character indicating alphabetic, numeric, control character, etc.
It's probably even quicker to check a character that way than the range
check. Something like:
unsigned char c;
if( table[c]&'\x01' ) printf ("alpha");
[C isn't my first language, but you get the idea.]
jmfbahciv
and now do the code in machine language :-).
/BAH
Peter Flass
Looking through the old manuals is fun! Every computer manufacturer
seemed to have their own unique version of BCD.
jmfbahciv
It wasn't nice in aulden days, especially if the table caused a page
fault.
/BAH
jmfbahciv
<snip>
To answer the question: SIXBIT. ;-) For the rest, you punt and leave it
as an exercise for the customer because anything you decide to do
will be "wrong" for somebody.
/BAH
Joe Pfeiffer
> Joe Pfeiffer wrote:
>>
>> if (((char >= 'A') && (char <= 'Z')) ||
>> ((char >= 'a') && (char <= 'z')))
>> printf("alpha!");
>>
>> I notice you assumed a verify() function, and didn't implement it.
>>
>> Also, of course, I regard the very first principle of programming to be
>> "avoid circumstances that make bugs even easier" -- possibly a corollary
>> to the Act of Contrition, which states "I firmly resolve, with the help
>> of Thy grace, to sin no more and to avoid the near occasions of sin."
>> (yes, I am a Catholic).
>
> and now do the code in machine language :-).
And for your penance....
Quadibloc
> I certainly wouldn't check for alphabetic in any other way. Bit
> twiddling or range checking is fine, but how do you handle accented
> characters, which are also alphabetic? Using a table or, better yet,
> internationalization functions to determine alphabetic is the only safe
> way these days.
Yes, that is true these days. But for that case, one does not have to
argue the merits of ASCII over EBCDIC, since the only applicable code
is the Unicode derivative of ASCII.
John Savard
Scott Lurndal
>
>"Peter Flass" <Peter...@Yahoo.com> wrote in message
>news:h33b2a$8b4$1...@news.eternal-september.org...
>>
>> Never caused me any problems, and still doesn't. Why do you need them
>> contiguous?
>>
>> if verify(char,'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ')>0
>> then put('not an alphabetic character');
>
>Apart from the maintainability question (checking that the 52 characters are
>correct takes longer than just checking two - or do you just assume the
>original coder doesn't make typing errors?) this is much less efficient (in
>both its usage of memory and CPU time) than is possible on an ASCII system:
>
>IF ('A' < CAP(char)) OR (CAP(char) >'Z') THEN Out.String("not an alphabetic
>character") END;
Not even the standard C 'isalpha' is implemented this way. Pretty much
every implementation uses a 256 entry array and uses the character as an
index. Various bits represent 'alpha', 'numeric', 'whitespace', etc.
This, of course, doesn't require contiguous alpha characters.
scott