Re: [gentoo-user] [OT] Hardware Problems causing kernel panics during large compiles

[Michael Mol]

Jason Weisberger wrote:
> Here's the rig:
>
> AMD Phenom II X3 720 (unlocked to 4 cores and OC'd to 3 Ghz, however
> taking it back to stock doesn't affect the problem)
> 4 Gigs DDR3-1600 at 7-7-7-16
> ASUS ASUS M4A78T-E AM3 Mobo
> MSI N9800GT GeForce 9800 GT 512MB
>
> other less important things would be two 320GB SATA drives, an IDE dvd
> writer, internal card reader, blah blah
>
> There are two things I can get this system to consistently lock up
> while doing: a large compilation such as Chromium, open office, or
> GCC, or playing Age of Conan in my separate Windows 7 partition.  For
> normal everyday tasks, I can't lock the thing up.  The system even
> plays WoW on maximum settings, Skyrim on max, Counter-Strike Source on
> max, 1080p youtube or local videos / movies.
>
> Watching the processor temp stays relatively low, the GPU temp stays
> normal, north bridge temp is good.
>
> The kernel panics I get are usually recoverable, I can just F7 back
> into my desktop and the compile may or may not still be going.  If it
> is, it's usually dead by the third or fourth panic.
>
> Memtest86+ ran for 24 hours and found no issues with memory on 18 passes.
>
> So, pretty much where I'm at is either processor or RAM.  Unless
> anybody here has a better idea.  And how could I test whether it was
> the processor or RAM without any current replacement parts?  What
> tests usually show one over the other?

Chromium, at least, is going to force you to start swapping if you've
only got 4GB of RAM in the machine, so I'd check smartctl and see if the
drive you swap partition on is having any trouble.

I'd also suggest removing the NVidia card, going without X for a little
while[1], and see if you can reproduce the issue[2].

I'd guess that it's the unlocked core that's the source of your woes,
though. Not all X3s are so labeled just for market segmentation.

[1] Only because your onboard video is ATI, and flipping between ATI and
NVidia configurations for something like this would strike me as too
much a hassle.

[2] Only because limiting the number of active components helps when
troubleshooting hardware issues. I once found a sporadic kernel panic
was related to a failing NIC. Another time I found it was related to an
old Hauppauge PVR150 that worked fine with my old motherboard, but was
incompatible with my new motherboard.