More blue screens, reboots and lockups on a new A7N8XED

[Doug]

Hi All,

I posted a few weeks ago about some blue screen (0xa stop), spontaneous
reboot and lockup problems with a new A7N8XED system running Win2K. "Most"
of that problem seemed to correct itself after I removed references to old
Via drivers in the registry, but I continue to have sporadic blue screens,
reboots and lockups. I am suspecting something involving memory, since the
problems seem to disappear when I configure memory to not run in "dual
channel mode" (I'm using Infineon 3200 memory). The problem also seems to go
away if I reduce the speed of the memory.

I don't know if it's of any help, but the blue screens I'm getting are:
"KMODE exception not handled" -- ntoskrnl.exe 0x0000001e , and "Page fault
in nonpaged area" 0x00000050 win32k.sys . This morning, after seeing that
page fault blue screen, the machine would not successfully reboot by
pressing the restart button -- it would get to the Win2K. logo, and would
not lockup, but could not go any further. Additionally, as I say, the system
sometimes reboots by itself, and sometimes freezes entirely. Not too fun...

I'm thoroughly mystified here. If the memory sticks were bad, I would think
that they would also fail when not in dual channel mode. But perhaps dual
channel mode puts some additional strain on them. Or maybe it's the
motherboard, or CPU, or ?

Thanks in advance for any clues...

Doug

 

[Ben Pope]

Hi All,

I posted a few weeks ago about some blue screen (0xa stop), spontaneous
reboot and lockup problems with a new A7N8XED system running Win2K. "Most"
of that problem seemed to correct itself after I removed references to old
Via drivers in the registry, but I continue to have sporadic blue screens,
reboots and lockups. I am suspecting something involving memory, since the
problems seem to disappear when I configure memory to not run in "dual
channel mode" (I'm using Infineon 3200 memory). The problem also seems to
go away if I reduce the speed of the memory.

I don't know if it's of any help, but the blue screens I'm getting are:
"KMODE exception not handled" -- ntoskrnl.exe 0x0000001e , and "Page fault
in nonpaged area" 0x00000050 win32k.sys . This morning, after seeing that
page fault blue screen, the machine would not successfully reboot by
pressing the restart button -- it would get to the Win2K. logo, and would
not lockup, but could not go any further. Additionally, as I say, the
system sometimes reboots by itself, and sometimes freezes entirely. Not
too fun...

I'm thoroughly mystified here. If the memory sticks were bad, I would
think that they would also fail when not in dual channel mode. But
perhaps dual channel mode puts some additional strain on them. Or maybe
it's the motherboard, or CPU, or ?

Memory sounds like a good starting point... as well as CPU.

Grab memtest86 and check the RAM, if you get errors, try upping the memory
voltage 0.1V.

Whats the RAM?

Ben

 

[Paul]

Hi All,

I posted a few weeks ago about some blue screen (0xa stop), spontaneous
reboot and lockup problems with a new A7N8XED system running Win2K. "Most"
of that problem seemed to correct itself after I removed references to old
Via drivers in the registry, but I continue to have sporadic blue screens,
reboots and lockups. I am suspecting something involving memory, since the
problems seem to disappear when I configure memory to not run in "dual
channel mode" (I'm using Infineon 3200 memory). The problem also seems to go
away if I reduce the speed of the memory.

I don't know if it's of any help, but the blue screens I'm getting are:
"KMODE exception not handled" -- ntoskrnl.exe 0x0000001e , and "Page fault
in nonpaged area" 0x00000050 win32k.sys . This morning, after seeing that
page fault blue screen, the machine would not successfully reboot by
pressing the restart button -- it would get to the Win2K. logo, and would
not lockup, but could not go any further. Additionally, as I say, the system
sometimes reboots by itself, and sometimes freezes entirely. Not too fun...

I'm thoroughly mystified here. If the memory sticks were bad, I would think
that they would also fail when not in dual channel mode. But perhaps dual
channel mode puts some additional strain on them. Or maybe it's the
motherboard, or CPU, or ?

Thanks in advance for any clues...

Doug

The Nforce2 chipset has a bug in it, and for error free memory operation,
you should set the memory clock to be the same as the FSB. This
is referred to as running in sync (synchronous operation) or 100%.
Now, some chipsets benefit from running the memory faster than the
processor can consume the data, but the Nforce2 isn't one of them.
The additional latency from the chipset reclocking the data when
the signals change clock domains seems to negate any benefit from
a higher memory clock. So, you should be running in sync.

Looking in the other threads you've created, you already know about
adjusting the multiplier and FSB, to get a higher FSB/memory clock,
while maintaining the processor core at roughly its rated
frequency. This works on the older processors. If you buy a processor
now, a different locking mechanism is used in the new ones, which
only leaves increasing the FSB as an option. The result is the core
operates higher than its rated speed. (I've seen a couple of posts
about a software mechanism to unlock the processor, but haven't seen
any posts about that recently.)

As far as the cut bridges on the top of your processor is concerned,
can't you see which ones are cut with a magnifying glass ? A little
conductive paint, like a car window defroster repair kit, could be used
to reconnect whatever you've cut. (After all, repairing an already
unlocked processor will be a lot better than buying a new locked
processor that doesn't have a hardware hack to unlock it.) There
are a number of web sites you can find with a search engine, with
instructions on filling pits (to avoid shorting to the copper
in the bottom of the pit), and then connecting the gold dots with
comductive material.

HTH,
Paul

 

[Doug]

Thanks Paul, I guess you've probably solved the mystery. I was trying to run
the memory at 200 with the fsb at 133 or 166. I ran memtest this morning (at
133) and found no problem (thanks Ben and rstlne). My next step was to do as
you suggest, and reconnect the cut traces. I don't know why the cut traces
prevent me from running the cpu at its rated speed, but hopefully it will
behave itself with the traces back in place. I can then try the memory in
dual mode at 166mhz, with the cpu at 11X166, and, God willing, live happily
ever after, more or less, at least until I'm ready to buy another cpu..

Doug

 

[Doug]

Paul, you really saved my life, it seems. I've had no more problems since
setting the memory at 166. I remember trying that in the earlier stages of
troubleshooting, but at that point there were evidently other problems
causing the crashes as well.

Such a simple detail, but in all my searching, at the Asus web site and
other locations, I found no mention of it. Of course I'm sure it's been
mentioned on this newsgroup, but things are sometimes hard to find in such a
forum, and articles expire.

I've experienced this with other motherboard manufacturers as well. The
owner's manuals give absolutely no advice about how to set all those dozens
of options in BIOS, in fact the descriptions of the settings often are
meaningless unless you're already knowledgeable about the motherboard and
chipset. It would seem so useful to have a paper somewhere detailing the
function of each setting, possible advantages and pitfalls to changing the
various settings, and most importantly, known bugs such as the one that has
kept me in computer hell for the last month. God I'm so happy not to see any
blue screens for a few days! ;-)

Anyway, I probably won't even bother reconnecting the traces at this point.
I'm running at 13X166, which I guess is the speed for a 3000+ (that's what
the CPU is reported as). Someone suggested that that was as much
overclocking as I should attempt for my 2500+, and I'm guessing that the
overall performance is comparable to what I'd get running at 200 MHz with a
lower multiplier (if 200 MHz was even possible for a barton 2500+).

Thanks again, everyone. The computer has returned from the land of the dead!

Doug