Card Columns (was Why did they make ... ?) 
Author Message
 Card Columns (was Why did they make ... ?)



Quote:




>> You're missing a very important point; sort is a dual
>operation.
>> Without a lot of fussing, you don't compare three things at a
>time.

>> Size of memory means nothing other than an immediate access to
>> the next item to be sorted.

>True. You can sort an enormous amount of data with very little
>memory
>and with external sequential-access devices, by using merge
>techniques.

>That's how it was possible for an insurance company or bank to
>perform all
>its accounting using machines with only 16K of memory back in the
>'60s.

>That method is simple and efficient ... it's always O(n*log(n)),
>and with
>today's abundance of RAM, can even be done without tapes or
>disks, almost
>all the time.

>What point did I miss, though? I don't follow you.

I'm not arguing about your algorithm; the others are know
more about this stuff than I do.  What I'm trying to point
out, and doing it badly, is that there is more to designing
a computing task than capacity or speed.  

/BAH

Subtract a hundred and four for e-mail.



Thu, 07 Jul 2005 20:01:06 GMT  
 Card Columns (was Why did they make ... ?)

Quote:
> I'm not arguing about your algorithm; the others are know
> more about this stuff than I do.  What I'm trying to point
> out, and doing it badly, is that there is more to designing
> a computing task than capacity or speed.

You're absolutely correct. I couldn't agree more.

However, this particular discussion happens to be
about the issues of speed and capacity, which, I
admit only become important when an application
is designed that is too limited in capacity, or when
it runs too slowly to meet the user's expectations.

Believe me, after 36 years in this business, I've found
a lot more to worry about than speed and capacity,
but I can't ignore them either.

If you find this discussion boring, perhaps you'd prefer
to read some of the other threads instead.

To tell the truth, I'm getting bored with it myself. I've
come to realize it's now starting to get down to a
competition to see who has the last word, and that leads
to an infinite loop.

Like many other things in the world, there's no such thing
as a "perfect" sort algorithm.



Thu, 07 Jul 2005 23:42:17 GMT  
 Card Columns (was Why did they make ... ?)

Quote:
> On Sun, 19 Jan 2003 03:02:10 GMT, robin


Quote:

> <snip>

> >Probably false advertising, as didn't IBM have a 1401 emulator
for the
> >/360 ?

> More of a microcode assist for a simulator.  Mod 30 & 65 I am
sure of,
> and I think 40 & 50 as well.

> I know of someone who rewrote the mod 30 program, so that it
could run
> two 1401 sessions at once.

1401 emulation was available on the model 30 and 40. The higher
models had
available emulation for other machines in second-generation
product line, such
as the 1410, 7010, 7070, 7090, 7094, and perhaps a few others.

The 1401 mode on the 30, I believe, was actually "compatibilty"
rather than
emulation as such, as it could run entirely without a
360-architecture program.
That is, it wasn't just an "assist".  On the 40, it was a
combination of microcode
and 360-mode operations.

With any of the emulators, however, there was far more
microcode-level
execution than 360-mode stuff. As far as I know, with any of
them, they only
dropped out to the 360-level to perform I/O. This was a good
thing, because it allowed
the emulators to co-exist with an operating system, although
initially they weren't
written that way.

As to performance, the 1401 mode on both the model 30 and the 40
outperformed the
original 1401 itself. I gather that the Honeywell system being
offered (the H200?) also
outperformed the 1401, but didn't stack up to what the 360's did.
Expecially when
applications were ported to run natively in 360-mode. So
Honeywell was offering
a debatable "easy migration" into a non-expandable architecture.



Fri, 08 Jul 2005 00:07:59 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


> >       [snip...]          [snip...]           [snip...]

> > Not all machines used punch cards -- many used paper tapes
> > instead of cards! The tapes were prepped on teleprinters. See
> > BAH's posts -- she started in DEC in TAPE PREP -- not keypunch.
> > Teleprinters were developed for telecommunications and paper
> > tapes used for message switching globally before card accounting
> > became popular. There's only a small step required from feeding
> > paper tape into machines to having teleprinters directly
> > connected to machines.

> In Steven Levy's book _Hackers_, he writes about the early
> hackers at MIT using the TX-0 and PDP-1 computers with paper
> tapes. If you wanted to assemble your code, you loaded the
> assembler binary program from paper tape. Then you input your
> assembly source code via another paper tape. The "T" at the
> end of the FLIT de{*filter*} name, and the "T" in TECO both
> originally meant tape, as in paper tape. And "DDT" was the
> "dynamic debugging tape". IIRC.

Paper tape use was popular from "day 1" of computers
as the technology existed prior to computers (as was
punch cards for the same reason).  It also remained
popular up until about 1980 or so.  In the business I'm
in I'm replacing a number of systems installed in the
70's and many of them have paper tape readers as
part of them.  Their use was common on many mini
computer systems.  Further, NC machines were
frequently programmed with paper tape well into the
80's and may still be in some places but I haven't had
any contact with the NC world in over a decade so I
cannot be sure.  Early hobby computing distributed
software on paper tape although it was not popular
as the readers were expensive.  This lead to a number
of inexpensive "pull it through yourself" "home" paper
tape readers.  I have one from RAECO, another
company OAE (Oliver Audio Engineering IIRC)
advertised one widely too.  I used mine frequently for
a couple of years in the late 70's.  The Kansas City
Audio tape mechanism, since it was cheap and
allowed both reading and writing replaced it though
which in turn was replaced with the floppy disk.
Although there were "home" paper tape readers,
"home" paper tape punches were beyond inexpensive
implementation.  I didn't get a "very used" paper tape
punch until I acquired an old ASR33 in the mid 80's.

BTW, I do agree with the poster that the 36 bit
machine had something to do with the 72 column
card limit but I seem to recall the 407 (??) accounting
machine only had 72 columns of print out as well
and maybe it impacted this limit as well.  This
accounting machine preceded the 700/7000 series
of IBM computers.  Recall that the UNIVAC cards
had 90 columns.

Chris
AN GETTO$;DUMP;RUN,ALGOL,TAPE
$$



Fri, 08 Jul 2005 02:39:55 GMT  
 Card Columns (was Why did they make ... ?)
On Sat, 18 Jan 2003 15:52:44 -0700, "Russ Holsclaw"

Quote:



>> I suggest that you examine your assumption w.r.t. the newsgroup
>> alt.folklore.computers.

>Ahh! I hadn't noticed that this thread was cross-posted to both
>a.f.c. and the PL/I folks.

>Anyway, I said "many" people in the newsgroup, not "most".  I'm
>following this via a.f.c., not c.l.pl1 anyway.  The remark was in
>reference to those who were using the 72 columns of TTY machines
>as the origin of the practice of
>reserving 73-80 on punched cards for sequence numbers. I still
>hold that the TTY assertion is not the correct  explanation, for
>reasons I gave earlier, related to the 704/709/709x card readers
>being unable to read more than 72 columns.

>In saying this, I'm merely asserting that the conventions
>established for punched card source files in programming
>languages had a punched-card origin not a TTY origin.

>I didn't intend to ruffle any feathers here. Minis have a
>venerable tradition, too. It just happens that their systems
>reflect their TTY-oriented heritage in just the same way as
>mainframes (and most particularly IBM ones) reflect their
>punched-card heritage.

I'd say that IBM mainframes reflect their punch card heritage,
while other manufacturers' mainframes provided support for both
punch cards and paper tape, having neither a Hollerith nor a
telecomms heritage.

Quote:
>With respect to PL/I, I recall that it was developed by a team in
>the UK, I also recall that, in the late '60s, anyway, the PL/I
>compiler for OS/360 was the only one IBM provided that accepted
>variable-length source files in addition to fixed-length files
>with 80-byte record lengths.  All the rest supported punched-card
>format only.

>I'd be interested in hearing some historical background as to why
>paper tape was more popular in Europe instead of cards.

IMO the administration and commercial interests of the major
European nations and their far flung imperial colonies required
high speed communications which developed using paper tape over
long haul cable and then wireless circuits, so there was already
a government and business infrastructure which had developed,
used, and "understood" paper tape for processing information.
And, as in more recent decades, up to the present, if it came
from the USA, and especially if a technology seemed to be a
single company monopoly, one things the Europeans always seemed
to be able to agree on, was to use something different, or even
invent something to be different, to protect their industries.
Nowadays, with the European Union having its common standards,
the shoe seems to be on the other foot, and it seems to be the
USA who is determined to go off in some other direction, to
protect its industries.
Of course, the Asians just create a common design with better
features and a couple of minor variations, and outcompete both
continents' industries.

Thanks. Take care, Brian Inglis         Calgary, Alberta, Canada
--

    fake address                use address above to reply






Fri, 08 Jul 2005 02:55:13 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> I'm not arguing about your algorithm; the others are know
> more about this stuff than I do.  What I'm trying to point
> out, and doing it badly, is that there is more to designing
> a computing task than capacity or speed.  

And sometimes you have to fight hardware or OS's or a venders
software library to get a job done.
Ben.


Fri, 08 Jul 2005 03:27:26 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> I didn't intend to ruffle any feathers here. Minis have a
> venerable tradition, too. It just happens that their systems
> reflect their TTY-oriented heritage in just the same way as
> mainframes (and most particularly IBM ones) reflect their
> punched-card heritage.

UR configurations (printer, card reader, controller, etc) were on the
order of the combined cost of mini plus TTYs. one might claim that
minis were TTY-oriented because of cost issues.

7094/ctss had 2741 support ... as well as cp/67.

at the university i implemented tty support in cp/67 ... with a peculiar
characteristic that later resulted in the following bug story (some
amount of the code i did at the university was picked up and distributed
as part of the standard product):
http://www.multicians.org/thvv/360-67.html
at:
http://www.multicians.org/thvv/tvv-home.html#stories

as part of the TTY support ... I attempted to extended dynamic
terminal identification using the SAD command in the 2702 ... being
able to change different line-scanner. 2702 short-coming was that
while it was possible to dynamically associate different line-scanners
with each line (both hard-wired & dial-up lines) ... they took a short
cut and hired wired the oscillator to each line (could switch between
the 2741 line-scanner and the tty line-scanner on the same line
... but couldn't dynamically change the baud rate).

this led to a project at the university to build our own controller
(supporting automatic terminal identification and automatic baud rate
identification) starting with reverse engineering the 360 channel
interface and putting it into an interdata/3 ... then extended to an
interdata/4 using interdata/3s as line-scanners. we since got blaimed
for originating the 360 pcm controller business:
http://www.garlic.com/~lynn/subtopic.html#360pcm

later it was bought up and the boxes were sold under the perkin/elmer
label.

in effect ... each of the (many) controllers in a typical mainframe
operation tended to be full-fledge minis in their own right

--

Internet trivia, 20th anniv: http://www.garlic.com/~lynn/rfcietff.htm



Fri, 08 Jul 2005 03:30:10 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> BTW, I do agree with the poster that the 36 bit
> machine had something to do with the 72 column
> card limit but I seem to recall the 407 (??) accounting
> machine only had 72 columns of print out as well
> and maybe it impacted this limit as well.  This
> accounting machine preceded the 700/7000 series
> of IBM computers.  Recall that the UNIVAC cards
> had 90 columns.

I belive the fact that 72 columns for fortran was
do to the fact card reader on the 704/709? could only
read 72 columns. I think this was done this way
so they could punch 36 bit words as 72 bit binary
data.


Fri, 08 Jul 2005 03:37:20 GMT  
 Card Columns (was Why did they make ... ?)
On Sat, 18 Jan 2003 20:44:43 GMT, "John W. Kennedy"

Quote:


>> Punch cards were invented by Hollerith in 1880, but I don't
>> believe they were in wide use after the 1890 census: mainly used
>> for statistics, so probably large governments only.

>Not at all.  Punch-card accounting systems were adopted in the early
>20th century just as quickly and enthusiastically as computers were
>adopted fifty years later.  IBM was already one of the biggest companies
>in the US before WW2, and the famous 1954 consent decree that hobbled
>IBM's business practices for so long was based on their 90% control of
>that industry, not on the fledgling computer side.

Maybe in the US, but the UK seemed to use a variety of different
equipment, and IBM gear and punch cards seemed to be used only by
the biggest government departments, utilities, and companies; at
least until the 360 came out and everyone produced their own
clone, at which point punch cards seemed to become popular.
Maybe their use was more widespread, but just not mentioned much
outside of their immediate areas of application: others from
rightpondia may have a different perspective?  

Thanks. Take care, Brian Inglis         Calgary, Alberta, Canada
--

    fake address                use address above to reply






Fri, 08 Jul 2005 06:34:37 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


>>I'd be interested in hearing some historical background as to why
>>paper tape was more popular in Europe instead of cards.

>I wasn't around at the time, but I suspect that as tape was used extensively
>in the communications sphere, when they were developing Collosus[1] and the
>early computers it was natural for them to use the machinery they already
>had available.  With EDSAC program tapes were punched using standard Telex-
>type perforators[2].   Later computers probably just followed established
>practice.

Very likely true.

It is probably relevant that paper tape was one heck of a lot cheaper
than punch cards.
--

                   Forsan et haec olim meminisse juvabit.
              Charter Member of the Vast Right Wing {*filter*}



Fri, 08 Jul 2005 07:12:55 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> > I have an old Teletype Model 15 sitting in my garage, vintage WW
> > II. And it has exactly 72 print columns. It also has a very
> > limited 5-bit character set that doesn't include things like '=',
> > '+', or '*'.  It would be hard to do FORTRAN with that, much less
> > PL/I or C.  The ASCII TTY didn't appear until the mid-'60s.


Quote:
> Nevertheless they did it:

>   & was used for +            - stood for itself
>   * was used for multiply     / for divide

>   % for left parenthesis      {losenge} for right parenthesis

I think you're mixing up two different sets of character substitutions.
The ones you list above are the "BCD duals" -- the code points that
corresponded to different glyphs on IBM "commercial" equipment than
on "scientific" equipment.

Pre-ASCII Teletypes already had the apostrophe and parentheses
characters so there was no need to substitute other characters
for these.  What they lacked (at least on most US equipment,
since there were several punctuation variants in use) were the
plus, equals, and asterisk.  The SDC time sharing system used
these substitution conventions:

  & was used for +
  : was used for =
  # was used for *

and apparently this was sufficient to make it possible to use
the JOVIAL language from a Model 28 Teletype.  The same set
of mappings were also used on MIT's Compatible Time Sharing
System, where the MAD programming language was used.

Some of these survived even into the 1970s: one of the BASIC
implementations in the first issue of Dr. Dobb's Journal was
designed for a pre-ASCII Teletype and used the colon for the
assignment operator.

Eric



Fri, 08 Jul 2005 07:17:43 GMT  
 Card Columns (was Why did they make ... ?)

[... I delete much ...]

Quote:
>BTW, I do agree with the poster that the 36 bit
>machine had something to do with the 72 column
>card limit but I seem to recall the 407 (??) accounting
>machine only had 72 columns of print out as well
>and maybe it impacted this limit as well.  This
>accounting machine preceded the 700/7000 series
>of IBM computers.  Recall that the UNIVAC cards
>had 90 columns.

I think, but I do not know, that the IBM 407 could read all 80 columns;
at least, the one I used could.  The machine was programmed by a plug
board.  In the one I had access to, the plugs for columns 73-80 were
wired to a dead socket.  Column 1 was wired, in a way I never
understood, to throw a page if it saw a '1', to overprint if it saw a
'+', and do something unusually magic for other Fortran characters.

Unfortunately, some vandal unplugged everything one day.

The UNIVAC cards had round holes.  Years later, IBM produced a 96 column
card with smaller round holes.

Quote:
>Chris
>AN GETTO$;DUMP;RUN,ALGOL,TAPE
>$$

--

                   Forsan et haec olim meminisse juvabit.
              Charter Member of the Vast Right Wing {*filter*}


Fri, 08 Jul 2005 07:24:29 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> BTW, I do agree with the poster that the 36 bit machine had something
> to do with the 72 column card limit but I seem to recall the 407 (??)
> accounting machine only had 72 columns of print out as well and maybe
> it impacted this limit as well.

I agree that the 72-column limit was because of the 704's hardware
limitations.

The 407 manual says that it has 120, not 72, print wheels, so I don't
think it can be blamed.  I wonder what they did with all those columns
when the cards only held 80 characters apiece.

Eric



Fri, 08 Jul 2005 07:37:53 GMT  
 
 [ 373 post ]  Go to page: [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]

 Relevant Pages 
 

 
Powered by phpBB® Forum Software