Memory or Motherboard?

[Mikep]

I have a gateway core 2 thats about 2 years old with an Intel Love Valley
MB. About once every 2 or 3 weeks it will start up with the 'Windows
recovered from a serious error.... bla bla". Then I run Memtest and, sure
enough, it finds memory errors (usually with the same bit -0x02000000).

A temporary fix is to pop out the memory sticks ( 2 @ 1gb) and move them to
the other memory slots (white clips to black clips or the reverse). It runs
fine for another 2 to 3 weeks and then repeats....

I'm guessing it's the memory because the problem exists even when the memory
moves --- but a couple other opinions would be valuable before I spring for
new memory.

TIA,

M

 

[larry moe 'n curly]

I have a gateway core 2 thats about 2 years old with an Intel Love Valley
MB. About once every 2 or 3 weeks it will start up with the 'Windows
recovered from a serious error.... bla bla". Then I run Memtest and, sure
enough, it finds memory errors (usually with the same bit -0x02000000).

A temporary fix is to pop out the memory sticks ( 2 @ 1gb) and move them to
the other memory slots (white clips to black clips or the reverse). It runs
fine for another 2 to 3 weeks and then repeats....

That's not a temporary fix; the error just doesn't occur very often to
show up during normal use. Try running MemTest86 or MemTest+ every
day, and there's a good chance the error will be seen daily. BTW if
even MemTest finds the error, then it's really, really serious because
that's a much worse diagnostic than MemTest86 or MemTest+.
RealWorldTech.com reviewed diagnostics, several years ago, and found
wide variations in sensitivity.

I'm guessing it's the memory because the problem exists even when the memory
moves --- but a couple other opinions would be valuable before I spring for
new memory.

If the location of the error changes address when you move the memory
modules to other sockets, then the problem is definitely with the
memory. Otherwise suspect a power problem, either with the main power
supply (if you're lucky, it's just a loose or dirty connection) or
with one of the voltage regulators on the mobo, such as the one that
feeds 1.8V for the DDR2 sockets (or 2.5V for DDR modules). Bad
electrolytic capacitors are famous for causing voltage problems, but
major brand computers tend to use brands of capacitors that fail a
lot, lot less than the cheapo Taiwanese and Chinese stuff. However a
few years ago, Dell had a lot of failures with Japanese Nichicon
capacitors, normally one of the most reliable brands.

 

[larry moe 'n curly]

(to clarify)

... but the error addresses stayed the same when the modules
weren't moved...


Incompatible motherboard bios can also cause problems or
just a defective motherboard design that can't make full use
of all memory slots at the spec'd speeds, though this is
less common today than in the past.

I forgot all about that, even though I have two mobos with such
problems

 

[Gerard Bok]

I have a gateway core 2 thats about 2 years old with an Intel Love Valley
MB. About once every 2 or 3 weeks it will start up with the 'Windows
recovered from a serious error.... bla bla". Then I run Memtest and, sure
enough, it finds memory errors (usually with the same bit -0x02000000).

A temporary fix is to pop out the memory sticks ( 2 @ 1gb) and move them to
the other memory slots (white clips to black clips or the reverse). It runs
fine for another 2 to 3 weeks and then repeats....

I'm guessing it's the memory because the problem exists even when the memory
moves --- but a couple other opinions would be valuable before I spring for
new memory.

Neither memory nor motherboard would be my first suspect.

A once in a fortnight memory problem would be very hard to
pinpoint. But that's obviously not what you are dealing with.
In that case, whenever Windows hits an error, it would memtest
take another fortnight to catch the next error

Given that there seems to be some pairing in errors, I would look
at environmental parameters.
E.g. memory operating on the treshold of it's voltage range,
being affected by temporary out-of-band values.
(Either by unsufficient voltage supplied, temperature influences
or neighouring equipment.)

Does the board allow you to modify memory voltage and / or timing
?

 

[Mikep]

Neither memory nor motherboard would be my first suspect.

A once in a fortnight memory problem would be very hard to
pinpoint. But that's obviously not what you are dealing with.
In that case, whenever Windows hits an error, it would memtest
take another fortnight to catch the next error

A little more data - when Windows hits the error, running Memtest will find
a ton of errors immediatly. Before the errors, Memtest will run for hours
with no problems. I'm using MemtestX86+ Version 1.70.

Given that there seems to be some pairing in errors, I would look
at environmental parameters.
E.g. memory operating on the treshold of it's voltage range,
being affected by temporary out-of-band values.
(Either by unsufficient voltage supplied, temperature influences
or neighouring equipment.)

Does the board allow you to modify memory voltage and / or timing
?

I'll check on what parameters can be set on the board. -- I am using a UPS
which will occasionally flag a power error (once in 2 months or so), but
doesn't seem to be related.