Wednesday, February 14, 2007


Another trip back to line printers and mainframes

In the comments section of this entry, Maggie notes:

I used to work in the computer labs at UMD when I was a student, and I can still hear the printers' whine as they printed line after line. It was part of the background noise of the lab. There were always the annoying kids (like you, I guess ;-) who sat there and printed out pictures during the last week of the semester when everyone else was trying to do their work.
Well, I actually wasn't one of the annoying kids, in this case (though yes, certainly in others), but was one of the computer operators and programming consultants (while being a student). So I, too, got to listen to the whine, and cancel the student jobs if they were running at inappropriate times.

There was another student trick I forgot about until now, related to the line printers: Clever and impish students (oh no, not I) would print hundreds of lines of 133 plus signs. The first one caused an overstrike, and the repeated overstriking of full lines of “+” would soon tear the paper, requiring that the operator suspend the current print job, reload the paper, and then cancel the suspended job before it continued printing and ripped the paper again. Quite the delight.

Maggie's comment continues:

I also remember our mainframe was so slow that at the end of the semester, we'd often get timed out of our connection while our programs compiled.
We had an IBM 370/165, which was state-of-the-art at the time, but seems laughable now. It had 1 MB of main memory. And I think, now, about the power of the thing sitting on my lap now, with 2000 times the RAM, and a 2+ GHz clock speed. I recently recompiled, as a test, a software package that we wrote between 10 and 15 years ago, in C++. It used to take a little over two hours to compile the whole package, on a 33 MHz Pentium II processor with 64 MB of RAM. I can't imagine how long it would have taken on the 370/165. Now, on this laptop, the whole thing compiles in three minutes.

During the last two weeks of classes at the end of each (13-week) quarter, turnaround time for jobs on the 370 would be many hours. Professors gave out the final assignments with five or six weeks to do them, but almost everyone waited for the last week or two... and found it impossible, as they were only able to get two runs in per day! They'd cry and whine, and beg the professor for an extension. Invariably, the extension would be granted.

That vexed me, because I'd do the assignment as soon as it was given and I had the information I needed in order to get it done. With turnaround time of just a minute or two, I could debug the program and get it working in just a day, three or four weeks before it was due. It kept my stress level down, but it was annoying to me that the slackers got the extra time.


Maggie said...

And the slackers of yesterday are the ATM programmers of today. ;-)

Barry Leiba said...

I hadn't thought of it that way.