P
Paul
I have just re-built my system with a A8V deluxe mainboard and Athlon 64
3500+ CPU and I have one weird problem that I have not seen before. The
video output will suddenly die and then after a few seconds the system will
just reboot. This happens when resizing a window or moving an app. This has
only happens once or twice a day and other than that the system is stable. I
do believe that it is video related as one time my monitor displayed a
frequency out of range warning when it happened.
I did a clean OS install and have loaded all the latest hardware drivers and
A8V bios. I am not over clocking in any way. Below are my system specs, any
help would be appreciated. I need to get the system stable before Doom 3.
(Just kidding, well sort of)
Asus A8V main board
Athlon 64 3500+
1 GB Corsair XMS DDR-3200 LL
ATI Radeon 9800 Pro
SB Audigy 2 ZS
Windows XP Pro
Thanks,
Lothar
Your post sounds similar to your situation. A video problem caused
by an underlying memory problem.
http://www.abxzone.com/forums/showthread.php?t=76172&highlight=a8v
"Try changing you async latency to 6 or higher for the memory timings
instead of auto. Change the trc to 7"
The Corsair will have the SPD coded for aggressive timings by default.
These timings may be too much for the A64.
There is also a page on the Mushkin site, claiming some timing
parameters are being too tight, "because the processor is so fast".
Another way of stating this, is the Athlon64 memory controller is
very complicated, and many many timing parameters are exposed. There
are more parameters than are specified in the SPD. On the
one hand, the designer at AMD who did this can say, "no matter what
bug is in the silicon, we have a workaround", but on the other hand,
the BIOS coders at AMI are probably getting heart failure from
trying to write code to set up the A64. It is possible that one
of the parameters is "off by 1", making the ram settings tighter
than the BIOS interface and timing numbers show.
(Look for "high tRCD", for a suggestion to bump up tRCD by 1.
BTW: This article needs a rewrite. And some substance to their claims,
like observing the timing with test equipment.)
http://www.mushkin.com/epages/Mushkin.storefront/EN/UserTemplate/27
So, as a start, try the advice in the Abxzone thread. Test the
results, using memtest86 from memtest.org as the first test. This
will require a blank floppy, or the preparation of an ISO CD for
booting, as memtest86 is a standalone code requiring no OS to run.
Run at least several passes of memtest86 error free, before concluding
your changes to the memory timings are good ones.
The second step, is to run Prime95 from mersenne.org. Use the
"torture test". This runs from your favorite Windows OS. This will
cause different access patterns than memtest86, and will be testing
the processor in a more "heated" condition.
The third step, is to try something 3D graphics intensive. I like
3DMark2001 running in demo mode, because I can leave it running
over night. If the computer hasn't rebooted by the next morning, it
is probably stable enough to use, and the testing can stop.
There are many other adjustments that can be made, that might make
a difference. But for the moment, I think your Corsair needs some
tuning, before you go any further.
HTH,
Paul