[ale] Bad Memory?

[Jeff Hubbs]

Danny -

If it were me, I'd pull the Ethernet and modem cards and try the exact 
same thing again to see if it made any difference.  If not, I'd start 
making some sort of RAM alteration.

"...it would always fail in a different place" is a symptom of thermal 
problems but it's not always a conclusive symptom.  If you did not mount 
the CPU and heatsink yourself, take it apart and check it for Homeristic 
assembly procedures, like what I encountered about a year ago when I 
found a heatsink with the backing paper still in place.

- Jeff

Danny Cox wrote:

> Okay, here's a stumper:
> 
> 	This weekend, I attempted to load RH 7.2 + XFS (from Cheapbytes) onto
> my wife's 1GHz Duron with 768 MB RAM.  I've already loaded this distro
> onto three other computers, a PIII, a 2X Athlon 1.2 GHz, and a 700 MHz
> Athlon, so the distribution has worked before.  Many folks on the xfs
> mailing list have used it too, so I'm confident in the distro.  I also
> tried just straight RH 7.2.  No luck.
> 
> 	I found and worked around one problem: when swap space is too large,
> anaconda aborts, overflowing an integer (grr).  Making a smaller swap
> space allowed me to get around this.  However, the kernel, 2.4.9-XFS was
> still oopsing all over the place.  I finally got to a login prompt,
> though, and copied over a 2.4.18-xfs kernel from my 700 MHz Athlon.
> 
> 	This kernel seemed much happier, so I copied over the source tree, and
> attempted to compile it.  Whoa!  Segmentation faults, internal compiler
> errors, and "unparsable command", followed by a LISPy looking construct,
> presumably the internal compiler code.  Hmm.  Memory problems, right? 
> This is further confirmed by the fact that it would always fail in a
> different place....
> 
> 	My quandry: already suspecting trouble, and having rsync die (and take
> gterm with it) during the /home backup previously on SuSE 7.1, I started
> memtest86 2.5, and let it run overnight.  It completed some 17 passes
> with zero errors....  Grr.
> 
> 	So, now what?  Assume memtest86 isn't finding the errors, and start
> swapping/removing DIMMs?  I've tried adding 'mem=256M' to the boot line,
> and I still get errors, so I assume it's the "lower" memory, but that
> may or may not be DIMM slot 0, no?
> 
> 	Ideas?  Other than, of course, a "boat anchor" sale ;-).
> 
> 	Thanks!
> 
> 	General stats:
> 
> 1GHz Duron
> 768 MB RAM (3 X 256; rated @ 133, but wouldn't POST, so back to 100)
> DFI motherboard (don't have model, but purchased within last 3 months
> from Monarch Computer) [VIA chipset (hmm)]
> Intel EtherPro 10/100
> USR/whoever they are now PCI modem
> NVidia GForce 2 GTS 64MB (no NVidia drivers, just XFree86)
> 
> 




---
This message has been sent through the ALE general discussion list.
See http://www.ale.org/mailing-lists.shtml for more info. Problems should be 
sent to listmaster at ale dot org.