oceanclub said:
I've been getting Error 132 memory-associated crashes in World of
Warcraft quite regularly, and I'm now pretty sure it's linked to
temperature. According to ASUS Probe, my stats as as follows:
* CPU Temperature: 47 C
* MB Temperature: 24C
Normally the MB temperature is lower, but if around this or 25C, WoW is
liable to crash. I presume the memory is heating up. The above
temperatures don't seem very high, and I already have a ThermalTake
Volcano 11 cooler on the CPU. Any ideas how to lower the MB/memory
temperature further?
Thanks,
P>
First of all, positively relating an error seen in Windows, to a
root cause, is a difficult thing to do. The bloat of Windows
will only serve to hinder the diagnosis.
I recommend two tests. Go to
http://www.memtest.org/ and get a
copy of memtest86+ . There is a version that will format and
prepare a bootable floppy for you. A second version is suitable
for use as an ISO CD image.
Memtest86+ is good for "hard faults" in the memory. If the memory
is actually defective from the retailer, that is the main value
of Memtest86+.
One problem with the Nforce2 chipset, is the silicon technology,
power/gnd to signal ratio are such, that operation at FSB400
with DDR400 RAM is not fully stable. Users call such chipsets
"picky about RAM", which means that not just any RAM will work
well. I learned this lesson the hard way, and it took me 1 week
of testing various combinations of sticks, in single versus dual
channel mode, with various ratios of FSB to memory and so on.
You could try going into the BIOS, and finding a setting that
tunes the memory clock. When that setting is 100%, it means the
same clock is used for the processor interface and the memory
DIMMs. So, FSB400 plus a 100% memory setting, gives DDR400.
FSB400 with 83% memory setting, gives DDR333 and so on.
Do not under any circumstances, select extreme values like 200%,
as damage could result.
So, you can turn down the memory, as a means of curing the
problem.
Bumping up the DIMM voltage supply, in the BIOS, can also
aid stability a bit. A value of 2.7 or 2.75V should be
safe with just about any RAM.
If you can get Memtest86+ to complete at least two full passes
of testing, then it is safe to go and boot Windows. Now, get
a copy of Prime95 from mersenne.org . That program has a
test, called "Torture Test". The test consists of a mathematical
calculation, which has a known answer. The test also uses memory,
and the combination of the pummelling of memory and CPU, happens
to be a good way of checking the stability of your Nforce2
chipset settings, the CPU, and the memory.
If you can do a number of hours of Prime95, without it reporting
"rounding errors" or other error types, _then_ go back to gaming.
I don't see a reason to suspect temperature has anything to do
with the problems here. Nforce2 has enough problems of its own,
without throwing temperature into the mix.
The cure for my memory problems on my A7N8X-E Deluxe, was to
buy Ballistix PC3200 2-2-2-5 memory. Runs stable as can be
now, but didn't when I used some old CAS3 generic RAM I had
lying around. An alternative, which I didn't use, was to turn
the memory way down. Cause that would suck...
You might also observe the +5V voltage level, with Asus Probe
or with MBM5 (mbm.livewiredev.com). Use one of those programs,
but not both at the same time. The A7N8X family uses +5V supply
extensively for powering processor and everything else. It
places a relatively heavy load on +5V, and not a lot of load
on +12V. If, when you game, or are using Prime95 torture test,
the +5V dips below +4.75, you might consider a power supply
with more beef on +5V (like say at least 25 amps current on
the +5V output). A 25 amp rating means you can handle a decent
AGP video card (like one of those video cards that has a Molex
power connector on it), as well as the motherboard.
HTH,
Paul
I've done all that already: memtest