Paul wrote:
Colin B. wrote:
Hey everyone;

I'm running XP SP3 on an older system. I had an E2200 CPU in an MSI
motherboard (P35 Neo2). Last night I dropped a new CPU in--an E8500.
Now the system will pass POST fine and the BIOS recognises the new CPU,
but Windows won't boot. It gets to the logo screen, then flashes a BSOD
and reboots.

I've gone into the advanced startup options and said "don't reboot on
failure" (or whatever the wording is), which allowed me to capture the
Stop code: 0x8e.

I ran Windows Memory test overnight and after 145 passes, it found no
errors so I think I'm safe there.

Also, I can't boot into safe mode. It does the same thing.

Ideas? Is it actually a bad CPU or Windows just being annoying?

Thanks,
Colin


Did you make sure the BIOS was updated to the suggested minimum version first ?

http://www.msi.com/product/mb/P35-Ne...div=CPUSupport

http://www.msi.com/product/mb/P35-Neo2.html#/?div=BIOS

I did. BIOS is at version 1C, higher than the 19 required.

Other than that, you're going from FSB800 to FSB1333, and if you were
previously overclocking, you might have a RAM divider set improperly
or something.

I've never overclocked this machine. Also, after some problems I went into
the BIOS and reset it with "load fail-safe defaults." The RAM divider is
set to "auto" and has clocked my RAM down to 800MHz.

It could be a bad CPU. That's certainly possible.

But good enough to pass POST, and load the XP installer from DVD? Weird.

You could go into the BIOS, record any custom settings necessary to
get the computer to boot. Then, shut down and "clear CMOS" and
enter the BIOS settings again. Choices such as "Auto" in the BIOS,
should select stock frequencies for things. The only thing you have
to watch, is that you get the boot order set up properly again,
and the disk interface in the right mode (IDE, AHCI, RAID etc).

That's a thought. I could definitely try clearing the CMOS completely.

Instead of testing with Windows, when you make major hardware
changes, you can instead test with a Linux LiveCD, as that
doesn't reference the hard drives at all. You can go to mersenne.org/freesoft
and get a copy of Prime95, and use that as a load test. There is a
Linux version, as well as a Windows version. So you can do some
basic stability testing, from Linux. But only do that, after
you've done some checking about BIOS version, cleared CMOS
(with all power removed from the system). Then, it should be
safe to set things up again, and do some testing from say
a Ubuntu LiveCD. If a LiveCD crashes, no one gets hurt. You
can leave the hard drives disconnected for safety while
doing that. That way, there is less likely to be some kind
of corruption on the hard drive.

That's the plan for tonight. I didn't have a livecd handy, and couldn't
download and burn one on my BSOD computer. :-)

The BIOS should be flashed, while your old stable processor
is inserted. Once the new BIOS is in place, you can install
the new processor. If your setup is at all unstable, it
generally isn't a good idea to flash the BIOS in that state.
Even with a dual BIOS motherboard, there is some danger involved.

Your "8e" error is "KERNEL_MODE_EXCEPTION_NOT_HANDLED". When
you get a BSOD, it's a good idea to write down the driver
name or the like, if it is printed on the screen. The kernel
can be tipped over by a driver.

There was none. It went from the preamble ("A problem has been detected..."
straight to "If this is the first time..." with no indications of what
caused the problem.

I wouldn't have thought your processor was so radically different,
that Windows is getting tripped up. Your E2200 is a dual core,
and so is the E8500, and they're both Core2 processors.
What's not to like ? I would think the same HAL driver would
work. Activation might be affected, if enough hardware changes
have accumulated, but you'd have already heard about that
from the computer, if that was the case. And activation
shouldn't be crashing the kernel. Something else is doing that.

Exactly what I was thinking. Technically the E2200 wasn't a Core2 CPU
though, it was a dual-core P4; but that distinction seems to be more
marketing than reality.

I'm assuming here, that your E8500 is brand new. If you got the
E8500 off Ebay, it could have been tortured by it's previous
owner, made to drink liquid nitrogen and the like. If it's brand
new, then you know it hasn't been abused. Batches of bad processors
do escape the fab, but not that often. In terms of the quality of
testing, they generally give the processors a good test before
shipping them. Cheaper parts ($0.25 ICs) don't get nearly as much
testing.

The E8500 is definitely not new, and may not be working properly. My
experience with CPUs has generally been binary, though - they either
work fine or not at all.

So this is my plan for the next round of "homework" (Bah, I'm too old to
be troubleshooting junk like this!):

1) Reset the CMOS and retry
2) Boot from a Linux LiveCD and test the CPU. I'm thinking of using
stresslinux - any thoughts?
3) Try to do a fresh XP install to a spare HDD I have.
4) ???

Oh, just in passing I should also mention temperatu At first it was
floating aroung 51C in the BIOS, but I decided to reseat the cooling
fan (used the original fan from the E2200--both CPUs are 65W, so it
should be sufficient), and now the BIOS is reporting about 41-43C; even
immediately after Windows crashes. If I can get the livecd to run, then
I should be able to monitor this closer.

Thanks,
Colin

--- Posted via news://freenews.netfront.net/ - Complaints to ---