[116949] in Cypherpunks
Re: CPU-Controlled Compromising Emanations
daemon@ATHENA.MIT.EDU (Dave Emery)
Tue Aug 24 08:58:52 1999
Date: Tue, 24 Aug 1999 02:38:02 -0400
From: Dave Emery <die@die.com>
To: Scott Guthery <sguthery@aa.net>
Cc: cypherpunks@toad.com
Message-ID: <19990824023801.A1014@die.com>
Mail-Followup-To: Scott Guthery <sguthery@aa.net>, cypherpunks@toad.com
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
In-Reply-To: <00c101beed92$40c3bb50$4c01a8c0@vagabond1>; from Scott Guthery on Mon, Aug 23, 1999 at 11:06:40AM -0700
Reply-To: Dave Emery <die@die.com>
On Mon, Aug 23, 1999 at 11:06:40AM -0700, Scott Guthery wrote:
>
> Play? Heck, that's how we debugged programs. You could
> listen to and count your loops. Vacuum tube machines such
> as the IBM 650 and PC machines such as the 1620 and 1401
> were all programmed with help of the the transistor radio debugging
> tool.
I've done that a little over the years (many years ago mostly),
but at DG in the 70s we really had a similar feature for some years
built into the old Novas. You see, the switching power supply on those
old machines switched at a low enough frequency so the steps in the
current flowing through the swinging inductor generated quite audible
magneticly coupled vibrations in the mounting hardware or the inductor
cores. This made the power supply sing or buzz at a high but audible
frequency modulated extensively by instantaneus changes in current
drain. And current drain changed dramaticly depending on where in
memory the cpu was accessing and what data was getting read or written
to the old ferrite core memory and how often the cpu was doing various
other things such as IO bus transfers. From this it was readily
possible (in fact hard to avoid) to hear the progress of programs as the
memory access patterns changed the current drain and the power supply
sang at a different pitch. One got very quickly to be able to identify
the progress of compilers, assemblers and linkers from the sound of the
machine, and one could sit back drinking a cup of coffee and tell just
how far one's build of a program had gone by listening to the power
supply singing pattern. And more dramatic, when debugging code it was
very often readily possible to actually hear a crash as a change from a
rough erratic sound as the code executed to a steady whine when the cpu
went into a tight loop or halted on the bug (or entered the debugger).
--
Dave Emery N1PRE, die@die.com DIE Consulting, Weston, Mass.
PGP fingerprint = 2047/4D7B08D1 DE 6E E1 CC 1F 1D 96 E2 5D 27 BD B0 24 88 C3 18