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


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

IBM's original cards had round holes too.  They switched to the
rectangular ones when they moved from 45 to 80 columns.

The story (true or false) is that they changed the shape because
they couldn't patent the number of columns they used, but they
could patent the superior structural integrity the rectangular
holes supposedly gave the cards.

Eric



Fri, 08 Jul 2005 07:41:15 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


>>There were a good many US manufacturers -- Burroughs, Sperry-Rand,
>>National Cash Register, Control Data, Honeywell, RCA, General Electric
>>(collectively known as the "seven dwarfs") -- even Philco made a stab at
>>it at one point, and had a fortran/ALGOL hybrid language called ALTAC.
>>But IBM held a commanding presence in the US business world almost from
>>the first, partly because they already dominated the pre-computer
>>punched-card-accounting business.  (IBM's machines had been entirely
>>programmable by the customer; Remington-Rand's machines had frequently
>>needed an engineer or even a trip to the factory.)

>>When the 360 came out, RCA simply copied its problem-state architecture
>>(though they had an incompatible supervisor state, requiring their own
>>operating systems), and Honeywell's entire advertising campaign, for
>>years, was on the theme, "We can move you from an IBM 1401 to a
>>Honeywell system more easily than IBM can move you to a 360."

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

>>--
>>John W. Kennedy

yeah but it was a HARDWARE based emulator which required that you re-IPL
the machine.
Software emulation didn't occur for quite a long while.....

/s/ bill turner, wb4alm



Fri, 08 Jul 2005 08:55:22 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> 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.

Spaces between fields.   The cards only had to be readable by operations
people, the listings frmo teh 407 by users.

// marc



Fri, 08 Jul 2005 09:32:00 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> 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

Well, yes and no.  The basic concept of the merge sort is simple, but
the execution is quite complex.  I know of three algorithms for tape
sorting, and three or four for disk.

--
John W. Kennedy
"The poor have sometimes objected to being governed badly;
the rich have always objected to being governed at all."
   -- G. K. Chesterton, "The Man Who Was Thursday"



Fri, 08 Jul 2005 11:53:22 GMT  
 Card Columns (was Why did they make ... ?)

Quote:


>>Honeywell's entire advertising campaign, for
>>years, was on the theme, "We can move you from an IBM 1401 to a
>>Honeywell system more easily than IBM can move you to a 360."
> Probably false advertising, as didn't IBM have a 1401 emulator for the
> /360 ?

Yes, mainly for the 360/30 and 360/40, where it was mostly implemented
in an optional extra microcode bank, but:

A) Originally, you had to boot part of the emulator from cards.

B) Later, it could run under DOS/360, but it still required disk volumes
dedicated to the 1401 file system.  Only cards and tapes could be used
in common.

C) And it left all your code in the original language, which was almost
certainly 1401 Autocoder (assembler).  Honeywell's "Liberator"
technology actually upgraded you to Honeywell native-mode operation (I
have no notion of how good "Liberator" was, but it must have been
decent, since, as I say, it was just about the _only_ feature of their
systems that they advertised, over a period of several years).

--
John W. Kennedy
"The poor have sometimes objected to being governed badly;
the rich have always objected to being governed at all."
   -- G. K. Chesterton, "The Man Who Was Thursday"



Fri, 08 Jul 2005 11:53:23 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> IMO systems where you don't need JCL are better designed for
> humans to use.  I never had a problem picking up and using JCL
> myself, but most application programmers did not have two clues
> about what numbers to plug in for dataset allocation, because
> they did not (want to) know device track sizes and record
> overhead counts, numbers of tracks and cylinders.

Whatever name you call it by, every system has some sort of more or less
rudimentary language that users employ to get their programs run and to specify
their input and output datasets.

Judicious use of cataloged and in stream procedures makes the need to explicitly
specify the SPACE parameter a rarity.  Moreover, if you don't like to measure
space in terms of cylinders or tracks, you can always measure it in terms of a
unit of your own choosing.

The features of the IBM mainframe operating systems that make a comprehensive
job control language necessary are the sources of their great strength.  These
include: 1) an IO system that is a comprehensive data management system which
includes a spectrum of data organizations and methods of processing instead of a
system that provides little more than low level transport and the association of
names with devices and (in the case of direct access devices) a chain of
allocation units. 2) a uniform, system-wide method of associating every input
and/or output requirement of a program with a suitable device, access method,
and data set.

In contrast, the widely used PC operating systems only provide builtin dynamic
binding for STDIN, STDOUT, and STDERR.  Moreover instead of looking on the
default binding as a basic fallback position for use when the user chooses not
to specify a binding, the default is presented as the standard from which the
user can choose to deviate by redirection which is described in a manner that
makes it seem almost magical.  While the net result is the same, this convoluted
nomenclature tends to produce a distorted mindset.

Moreover programmers were left entirely on their own for the remaining I/O
requirements of their products which has led to a confusing mishmash of
techniques, some reasonably satisfactory, others entirely unsatisfactory.
Consider the difficulty users have in getting two applications to coexist on the
same system if each requires a different version of some critical DLL.



Fri, 08 Jul 2005 12:34:37 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> 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.

Moreover IBM's practices that the government objected to were nowhere near as
predatory as Microsoft's.  Yet Microsoft is being handled with kid gloves
compared to the treatment IBM received.


Fri, 08 Jul 2005 14:09:40 GMT  
 Card Columns (was Why did they make ... ?)

Quote:



>>> Note that RPG was born on the 1401, a machine on which the COBOL
>>> compiler could take _hours_ to run.  RPG was the Turbo Pascal of its
>>> era.  It is also a very easy language to code, within its particular
>>> problem domain (which constituted a very sizable fraction of
>>> dollars-and-cents programming in those days).

>> Are you sure?  I started with IBM in June of 1960 in Cincinnati and I
>> don't recall seing a 1401 until sometime in 1961.  While I didn't use
>> it myself, I recall hearing about RPG from people who I thought were
>> working on 650's and 704's and/or 705's at the time.

> That's a little before my time (my first program was written in 1965 in
> Basic FORTRAN on Brown's 7070) but there are 1401 hardware artifacts in
> the RPG language, and the MIT Press "History of Early IBM Computers"
> says quite explicitly that RPG was developed for the 1401, and first
> delivered "early in 1961", "about three months" after the first 1401
> delivery.  And, strategically, the 1401 needed RPG as no prior machine had.

Well at least my memory of the date of the first appearance of the 1401 seems to
correct.  Perhaps the talk of RPG I remember hearing was by people who had been
sent on a course preparatory to the introduction of the 1401.  Of the dozens of
machines I have programmed in assembly/machine language I think the 1401 was the
most fun to program.  You could sure accomplish a lot in in very little space.
For example, in 1961/62 Bob Walton and I developed a stepwise multiple
regression program that ran on a 4K 1401 at the University of Cincinnati Medical
Computing Center.  It was a multi phase program. The first two phases processed
control cards and built up the correlation matrix (in triangular form) and the
right hand sides in the locations from 1000-3999.  For the critical third phase
that did the regression computations per se, we managed to pack a set of
variable presicion floating point routines and the stepwise regression
computation into locations 0-999, keeping the punch area available for punching
intermediate results to be processed later by the report writer and squeezing
the last little bit of the program into part of the read area.  I did the
floating point package and Bob did the actual matrix calculations.  We could do
a 25X25 problem with 7 digits of precision.


Fri, 08 Jul 2005 14:04:18 GMT  
 Card Columns (was Why did they make ... ?)

Quote:




>>>> Note that RPG was born on the 1401, a machine on which the COBOL
>>>> compiler could take _hours_ to run.  RPG was the Turbo Pascal of its
>>>> era.  It is also a very easy language to code, within its particular
>>>> problem domain (which constituted a very sizable fraction of
>>>> dollars-and-cents programming in those days).

>>> Are you sure?  I started with IBM in June of 1960 in Cincinnati and I
>>> don't recall seing a 1401 until sometime in 1961.  While I didn't use
>>> it myself, I recall hearing about RPG from people who I thought were
>>> working on 650's and 704's and/or 705's at the time.

>> That's a little before my time (my first program was written in 1965
>> in Basic FORTRAN on Brown's 7070) but there are 1401 hardware
>> artifacts in the RPG language, and the MIT Press "History of Early IBM
>> Computers" says quite explicitly that RPG was developed for the 1401,
>> and first delivered "early in 1961", "about three months" after the
>> first 1401 delivery.  And, strategically, the 1401 needed RPG as no
>> prior machine had.

> Well at least my memory of the date of the first appearance of the 1401
> seems to correct.  Perhaps the talk of RPG I remember hearing was by
> people who had been sent on a course preparatory to the introduction of
> the 1401.  

Of the dozens of machines I have programmed in

Quote:
> assembly/machine language I think the 1401 was the most fun to program.  
> You could sure accomplish a lot in in very little space. For example, in
> 1961/62 Bob Walton and I developed a stepwise multiple regression
> program that ran on a 4K 1401 at the University of Cincinnati Medical
> Computing Center.  

You bring back fond memories. My very first experience as a programmer
was on a 16K 1401 at North American Aviation in 1961. We used Autocoder
but later developed our own assembler because Autocoder took minutes to
spin through the program tape looking for macros even when our source
code had none.  For business data processing (decimal arithmetic and
string processing) the 1401 had (and still has) the best instruction set
in existence. In addition to the luxury of 16K, we also had 5 tape
drives, a card-reader/punch and a 300LPM chain printer. At some point we
added a paper tape punch to produce mylar tapes for some military
checkout equipment.  (Later, one of the major applications was rewritten
in a higher-level language for the 360-50 and it required over 120K !!!)

It was a multi phase program. The first two phases

- Show quoted text -

Quote:
> processed control cards and built up the correlation matrix (in
> triangular form) and the right hand sides in the locations from
> 1000-3999.  For the critical third phase that did the regression
> computations per se, we managed to pack a set of variable presicion
> floating point routines and the stepwise regression computation into
> locations 0-999, keeping the punch area available for punching
> intermediate results to be processed later by the report writer and
> squeezing the last little bit of the program into part of the read
> area.  I did the floating point package and Bob did the actual matrix
> calculations.  We could do a 25X25 problem with 7 digits of precision.



Fri, 08 Jul 2005 15:05:14 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> On later IBM sorters, it was possible to alphasort a file with only three
> passes per column. The trick was that the sorter would be set to sort
> 12-punch cards (A-I) into nine columns of the sorter. 11-punch cards (J-R)
> were all sorted into another pocket; 0-punch cards (S-Z) were sorted into
> yet another pocket.

> Pull out the A-I cards in order from 1-9, then sort the J-R cards on their
> numeric values and remove them in order; finally, sort the S-Z cards  into
> order and create one big deck.

That's not three passes, it's approximately 1-2/3.  The older method only took 2
passes.


Fri, 08 Jul 2005 15:39:08 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> Two other things.

> You could have it use column one as the carriage control character
> in printing the listing file.  This way you could make nicely
> formatted listings.  I even did one once with overprinting (+  CC
> character) to darken the
> major comment lines.   Also, make each procedure start a new page,
> things like that.

It didn't have to be column 1.  In the SM(l,r,c) option c=0 or omitted meant no
carriage control, any other value was the column that contained the carriage
control character which had to be either less than l or greater than r.


Fri, 08 Jul 2005 15:31:42 GMT  
 Card Columns (was Why did they make ... ?)

Quote:

> 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.

In 1967 I produced a film in collaboration with Ken Knowlton of Bell Labs.  We
used Ken's BEFLIX language whach ran on a 360 pretending it was a 7090 to
produce a magnetic tape which was then submitted to a Stromberg Datagraphics
4060 pretending it was a 4020.


Fri, 08 Jul 2005 15:55:39 GMT  
 Card Columns (was Why did they make ... ?)


Quote:
> 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.

IIRC, the abbreviation DDT stood for "dynamic debugging technique".

DDT *has* to have been a backronym.

--

    "We can no longer stand apart from Europe if we would.  Yet we are
    untrained to mix with our neighbours, or even talk to them".
                                              George Macaulay Trevelyan, 1919



Fri, 08 Jul 2005 16:06:39 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.

No, 407's had 120 columns.  So did 402's and 403's, but if I recall correctly
the 402 was numeric only and the 403 could print alphabetic information only in
the leftmost 40 or 50 columns.  Or maybe that was the 402 and the 403 had alpha
all the way across.  My memory is a little fuzzy on this point.

   This

Quote:
> accounting machine preceded the 700/7000 series
> of IBM computers.  Recall that the UNIVAC cards
> had 90 columns.

They really had 12 rows of 45 round holes.  The top 6 rows were columns 1-45 and
the bottom, columns 46-90.  They used an encoding scheme similar, if not
identical to, BA8421.


Fri, 08 Jul 2005 16:05:32 GMT  
 Card Columns (was Why did they make ... ?)


Quote:

> > 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.

> Moreover IBM's practices that the government objected to were nowhere near
as
> predatory as Microsoft's.  Yet Microsoft is being handled with kid gloves
> compared to the treatment IBM received.

Might be a case that IBM didn't pay their dues to the various people
and MS does.

Cheers,
Rupert



Fri, 08 Jul 2005 18:06:28 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