Re: woody/linux 2.4/raid/athlon/almost a horror story
On Sat, Jun 30, 2001 at 12:01:55PM -0700, aphro@portal.aphroland.org wrote:
> yeah i could try that, e2fsck was the only program i ran
> during that test boot, but im betting most everything
> else will cause the same error. do you have an athlon?
> do you run it with the athlon optimizations? maybe
> its a kernel bug..i read kernel traffic and have
> never heard of the athlon optimizations cpu option
I have tested compiling and running linux kernels on several
athlon-equipped machines and have not encountered any problems.
There were reports rumors of problems with newer via kt chipsets, in
some production series.
> i think the hardware is fine. its been running a week on
> 2.2.19 ive been stressing the opengl, upping system load to
> 10 for days on end running distributed.net, and running
> bonnie++ copying gigs and gigs of data to the filesystems to
> stress the I/O out ...not a single hint of an error.
Try compiling linux kernels. It is one of the best tests for the cpu
and cache. Do both single builds and parallel builds. If gcc starts
to crash with "signal 11" error, then that is bad.
Also check out va linux's cerberus (iirc).
Too bad your bios does not let you set bus and cpu speed, underclocking
the cpu would have been interesting to test. You cannot underclock
it either on the old motherboard? If you are afraid of filesystem
corruption, just disconnect the disks and boot from a floppy.
> again my power supply is a Power PC & Cooling Turbo
> Cool 450 ATX.
That could be good and it could be bad. I can't say a thing for sure
without having seen real measurements on it with an oscilloscope. And it
could have been the first one built on some monday morning. Have you
tried temporarily swapping it with a known good 300W power supply?
Does 2.4 work on that motherboard with a different athlon? With some
or most peripherals removed?
Have you tried compiling the athlon kernel on different machines, with
different gcc versions? Maybe you should start here, in fact.
Cheers,
Joost
Reply to: