Failing PowerPC system...

Don Dutile ddutile at redhat.com
Tue Mar 18 02:34:54 UTC 2008


Nothing like cleaning out the (electrostatic) dust bunnies
to make a computer run oh-so-much-better!

Congrats Jarod.... glad you cleaned that mess @home! ;-)


Jarod Wilson wrote:
> On Thursday 13 March 2008 05:44:30 pm Jarod Wilson wrote:
>> On Thursday 13 March 2008 10:46:49 am Jarod Wilson wrote:
>>> On Thursday 13 March 2008 12:23:18 am Dan Williams wrote:
>>>> On Thu, 2008-03-13 at 00:01 -0400, Jarod Wilson wrote:
>>> [OS X lockups and sha1sum failures under linux on ppc mac]
>>>
>>>>> So anyone have suggestions for diagnosing exactly what component is
>>>>> starting to fail? My first thought was cpu, then motherboard, and now
>>>>> memory... I figure memory is probably the easiest place to start
>>>>> poking, but is there a memtest86 equivalent for PowerPC? If not, I
>>>>> guess its memory stick roulette, and on to other hardware from there.
>>>>> Open to any and all suggestions...
>>>> RAM is definitely the first place to look.  You also might just want to
>>>> take the RAM out and firmly reseat it.  That's fixed these sorts of
>>>> things for me a few times before.
>> [...]
>>
>>> Just found http://pyropus.ca/software/memtester/ and will see if that
>>> builds/runs/does something useful under ppc32 linux or os x... There's
>>> also a non-free (but fairly cheap at $1.39) memtest os x at
>>> http://memtestosx.org/.
>> Never mind. Memory stick roulette suggests either all the sticks are bad,
>> or something else is failing. Backed off to running my sha1sum test with
>> only 1 stick at a time, and could reproduce the sha1sum failures on each of
>> them. Failures seem to happen more frequently after a number of iterations,
>> like after the system heats up a bit. On with the show...
> 
> So I finally yanked the mammoth heatsink and processor module, and noticed 
> that what I presume was once thermal paste between the cpus and the heat sink 
> looks a lot more like a thin layer of dirt than thermal paste. I cleaned off 
> both the cpus and heatsink, laid down some new thermal paste, buttoned 
> everything back up, and now 1000 consecutive sha1sums all return the same 
> result. Woo! Back in business...
> 




More information about the fedora-devel-list mailing list