20 Sep
2013
20 Sep
'13
1:52 p.m.
> Many computers read them that way as well, but
some readers move the
> card along the other axis, and so read a row of holes at a time....
On Fri,
20 Sep 2013, Chuck Guzis wrote:
I suppose one could make the argument that cards read
in row-binary
could be read faster than if read in column-binary, but all of the very
high-speed card readers that I've ever seen (e.g. CDC 405), read card in
column-binary fashion; i.e. 12 holes at a time, rather than 80.
The change from brush/brass-roller to light/photocell made a
difference there.
Brass roller reading had mechanical limits on speed, and 80 V 12
made a difference. With the change to photocells, SOME of the
mechanical speed limits were eased.
Half a century ago, I was hearing stories, . . .
S'posedly, IBM first tried patenting the shape of the hole. That
led to the existence of round-hole cards. But, that patent was
overturned.
Then IBM patented the brass roller. "Surely, THAT will stop CDC
dead in their tracks." Instead, not being able to use the existing
mechanical technology, they did the engineering to use photocells,
and leapfrogged IBM.
Our sources half a century were not authoritative, but it made a
good story, with a moral of "don't be too brutal with the competition,
it will come back to byte you.".
I suspect that may have been to accommodate
"short" cards (40 column)
more easily. The 405 had gizmos that allowed for short cards.
after most of the changes. I don't remember under-sized cards
being around when they sped up the readers. BUT, I was unaware
of what was going on in other circles.
When we talked about 40 column cards, we meant every-other-column
port-a-punch cards.
It was truly astonishingly impressive when IBM was able to
successfully read most of the CBS port-a-punch cards that
had a stamp stuck on them and dropped in USPS mailboxes!
It was less astonishing when IBM's data processing of the CBS
"National Driver's Test" resulted in percentage totals nowhere
near 100%. On the live broadcast, you could see Walter Cronkite
stalling while my father fed him manually recalculated results.
Starting the next week, our house was full of IBM FORTRAN manuals,
and FORTRAN textbooks, such as McCracken and Decima Anderson.
If they hadn't blown it, I might have become a mathematician
or a paleontologist.
In deference to Al's nit-picking, I think that the answer is that
while BULK QUANTITIES of blank cards for use in a punch could be,
and sometimes were, called "PUNCH cards", but that if the box has
been opened, then they should be assumed to be "punchED cards".
That would set a lower limit of 2000. Or should we apply the rule
only to the case of 5 boxes (10000 cards)?
And, of course, all of the IBM EAM equipment (there are some ads
on eBay) was called "PUNCHED card accounting systems", since what
would be the point of feeding blank cards into the read hoppers?
(makes the output of the 084 and 407 pretty predictable!)
--
Grumpy Ol' Fred cisin at xenosoft.com
http://www.xenosoft.com/FPUIB