Re: how can a bit be off in memory?
- From: "Robert M. Riches Jr." <spamtrap42@xxxxxxxxxxx>
- Date: Sat, 30 Jun 2007 21:24:11 GMT
On 2007-06-30, Jean-David Beyer <jeandavid8@xxxxxxxxxxx> wrote:
Robert M. Riches Jr. wrote:
On 2007-06-29, Charles T. Smith <cts.private@xxxxxxxxx> wrote:How would an alpha particle (helium nucleus) get through a computer case and
Vim started crashing on me, particularly when I tried to open new lines.
I finally checked it out with rpm and a newly downloaded copy of vim's
rpm and discovered that exactly one byte, deep into vim, was wrong.
I rebooted my machine (which has been super-solid for years) - and the
difference was gone.
So, what are the opinions - did I run into a hardware glich, or was
there a freaky issue with memory mapping?
If rebooting the machine made the (apparent) error go away,
then my guess would be a bit got flipped in RAM, the page of
RAM that held the cached copy of the page from disk. While
I worked at a large chip company, I heard of cosmic rays,
alpha particles, and such causing occasional soft errors in
RAM. (That's why ECC RAM was a primary factor in the
motherboard I chose when I built my machines.) I would
suggest running memtest86 overnight to check for something
more sinister.
through the plastic of a RAM chip? How could even a beta ray (electrons) get
through a metal case? The only thing I would expect to cause problems would
be gamma rays (cosmic rays). You might as well have ECC RAM (I do).
I agree it could easily be a cosmic ray (gamma ray or
something like a neutrino). However, alpha and beta
particles can be generated _INSIDE_ the machine, even inside
some chip packages.
There was a lot of discussion a few years ago about
radioactive lead in the solder used in the solder bumps used
to connect a chip with the conductors in the package. That
puts the particle (alpha, I think, but I might be wrong
there) _VERY_ close to the active part of the chip. The
particle only needs to go through the passivation and a few
layers of metalization and dielectric to get to the
transistors--on the order of a small number of microns.
I remember a few notable things about that discussion. One
report said IBM owned or had rights to a particular source
of lead ore with unusually low levels of radioactivity.
Everyone else (and IBM when that source ran out) sought out
lead that had been refined at least several decades ago.
The unstable isotope of lead would decay to things that were
either stable or had very short decay times. So, if you had
a block of lead refined a hundred or more years ago, there
would be almost nothing unstable in it. However, lead ore
would contain unstable isotopes of other elements that would
decay into the unstable isotope of lead. So, lead that was
refined less than a few decades ago would have significant
amounts of unstable lead atoms and could cause problems if
that lead were used in the solder on the bumps right next to
the chip. There were some reports that those seeking old
lead would go to buildings more than a few hundred years old
and freely replace roofs and/or pipes just to get the old
lead that would be safe for use in solder bumps.
--
Robert Riches
spamtrap42@xxxxxxxxxxx
(Yes, that is one of my email addresses.)
.
- Follow-Ups:
- Re: how can a bit be off in memory?
- From: Charles T. Smith
- Re: how can a bit be off in memory?
- Next by Date: Re: Latest spam from groups.google.com...
- Next by thread: Re: how can a bit be off in memory?
- Index(es):
Relevant Pages
|
