>> [...mysterious occasional coredumps on 1+...]

> The 'mrtg' machine has a quite high userland load: [...]

High load briefly every five minutes?  Hm.  My X sessions would have
three processes running once a second (as the clock ticks).  So
naïvely, your machine should last about 300 times as long (five minutes
divided by one second is 300).  And MTBF for my X session would be
somewhere around a week.

And a "make build" would keep the machine busy constantly, and it would
normally croak in less than a day.

> Now for sure I don't run X on the machine, it doesn't even have a
> framebuffer.

> Is the CPU in the IPC supposed to be the same as in the 1+ ?

I don't know.  I don't know whether IPCs exhibit this failure mode or
not; I've never had an opportunity to try it.  (For that matter, I
suspect the cache hardware more than the CPU proper, and I don't think
all the cache is on the CPU chip.  Hmmm, come to think of it, a few
cache lines is about the right size for the corruption I saw in that
picture.  I have a 1+ up, but idling, and dmesg says 16 bytes/line.  My
IPX shows 32 bytes/line, so they definitely have different cache
hardware - but it's not quite that simple; an ELC also shows 32.)

But this mrtg machine, it's not an IPC, is it?  I think the IPCs has a
framebuffer on the main board.  Or were you considering replacing it
with an IPC?

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B