On Saturday March 15 2003 09:02 pm, Ronald J. Hall wrote:
> Okay, I'm still having some hardware problems. I installed and ran
> memtest overnight, about 24 passes with no errors.
>
> I grabbed and installed cpuburn 1.4 last night, then ran burnMMX.
>
> It won't run longer than 2 minutes before I get a full lockup that
> requires a hardware reset or powerdown and back up.
Google: "Raising skinny elephants is utterly boring". There's
better ways to tryin shutdown before usin the panic button ;)
> That from a shell with X running and without.
Weak motherboard, power supply, ram, among other things including
bios (mis)configuration could be involved. Your mobo looks good but
that's a very poor memory vendor. Try setting the ram to cas3 with
bank interleaving disabled. Use lm_sensors and check frequently when
the system is under varying to heavy load to see if the voltages are
at or above spec and _stay steady_. See who made the PSU and check
to see if it, and the cpu heatsink (your mobo is) are AMD appr'vd
http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_869_4348,00.html
> This is an AMD XP2100, 512megs DDR (PNY) Ram, Soyo Dragon Plus MB,
> Nvidia Geforce 4 Ti4200, etc, etc, with nothing overclocked.
If you're using the proprietary nvidia driver, you should add
'mem=nopentium' to lilo.conf's append= line. Run 'lilo' and reboot.
Another very good system test is prime95.
ftp://mersenne.org/gimps/mprime2212.tar.gz
D/l and extract that file to a dir by itself. Then in that dir
run ./mprime -m and from the menu choose 17), the torture test.
You should be able to complete that without stopping on hardware
errors. Make sure you can before tryin cpuburn again. This mprime
test takes many hours, so plan on overnite.
> The CPU has a 5 1/4, 3000 rpm fan/heatsink attached to it, and BIOS
> reports 50 degrees celsius for it, and 27 degrees Celsius for the
> system.
Heatsink should have a 5000 rpm fan on it. The heatsink should
be sitting on the cpu die with a thin layer of thermal grease. A
thermal pad won't do, specially for Athlons.
Heat could well be causing your problems. The cpu temp you see in
bios is almost useless. Your system isn't fully booted into an OS
yet, and is under low load. To see if better heatsink or case
cooling is needed, take the case cover off and point a table fan into
the case. If your system then runs without, or with little problem,
you need a better hs-fan and/or case cooling. Do so soon, 'cause
letting the system overheat (too many times) will make the system
weaker from then on. Could even be permanently damaged (fubar'd).
> Intense games cause crashes, some with direct references to memory.
> Example, Warcraft 3 gives this:
Games don't make good system tests or benchmarks, despite many
people thinking they do. 'Sides, many crash all by themselves.
> The literature with cpuburn suggests that my CPU is not getting
> cooled enough but from what I've read, 50 degrees Celsius should be
> okay, right?
Yes, if it's monitored when the system is fully booted up and
working under load. No, when it's from bios. Keep in mind that the
temps you're seeing are from an external probe, and are only
indicative of approximate cpu temp. The cpu internal core temp is at
least 10C, if not as much as 20C higher. XP's made since last July
do have an internal diode to report core temp. Unfortunately there's
no desktop motherboards that implement it, or the very few that do,
at least not properly.
> PS Note that all the hardware is basically "new", less than 3
> months old.
Still, check for dust bunnies ... and make sure that cpu heatsink
isn't on a pad, and is on some grease. When you get cooling right,
you should be able to run cpuburn's burnK7 (or burnK6 will do) for at
least 15 to 20 minutes. Run for an hour, it means your system is
bulletproof. You might need to give that PNY ram to somebody you
don't like, and get some Crucial, Corsair, or Samsung cas2 ram ;>
--
Tom Brinkman Corpus Christi, Texas
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
