<html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body>
In Volume 2, Issue 8, Henry Baker comments on "listening" to old
mainframes using a radio:
<blockquote type="cite">
<pre class="moz-quote-pre" wrap=""><font size="+1">BTW, IBM mainframes had powerful radio side-channels: in the early
1960's, one of the programmers that worked with me used to put an
ordinary portable radio on top of the IBM CPU so that he could
"listen" to his program go through its paces. Any change in the
audio would indicate a problem with the software.</font></pre>
</blockquote>
In fact that was not an accident, in some cases it was a documented
feature. It was a part of other systems, but my experience with it
was with CDC 1604 and 3600 machinery. There was a speaker in the
operator's console, driven by the upper octal digit of the
accumulator. DC current proportional to that digit was fed through
the speaker. As an example, if you put 0707070707070707 (sixteen
digits octal, to fill the 48 bit register) in the A register and
then rotated it left or right, delay, rotate, delay, etc. you got a
square wave sound out of the speaker. More creative strings of
digits could create lots of different sounds. I remember a
diagnostic that CDC's onsite engineers ran that was called
"Bagpipe", because that is what it sounded like.<br>
<br>
I remember one programmer whose intense numeric calculations would
make the speaker go quiet, because the A register was changing at a
rate above human hearing during his intense computations. Operators
had learned that if programs in general shut up the speaker it meant
the machine was in a small tight loop, so they would interrupt the
run and kick his program off the machine. So the first thing his
code would do was to modify system code (this was before there was
much in the way of separation between user code and system code,
there was one big address space for us all to play in) and if the
operator changed a switch setting so as to kick him off, he printed
out on the console typewriter a message to keep their (text
suppressed in interests of decency) hands off the switch, and
telling the operator another way to stop the program if it were
really necessary.<br>
<br>
So if you consider that audio information as leakage, it was
endorsed by the system designers! Also, listening that way or using
a radio to listen in on activity with other machines was used to
play music, back in the day when we had to demonstrate the machine
to visitors.<br>
<br>
Bob W<br>
</body>
</html>