D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] Folding@home 'errors'

 

Dan James wrote:
On Tue, 2010-03-16 at 12:44 +0000, tom wrote:
You may have missed the point - f@h runs all the time on my system*, it only uses a small amount of the resources available but goes wrong from time to time - while the rest of the system doesnt. Basically the machine is running 100% all the time. * I've been running cpuburn flat out for about an hour now and there's no noticeable temperature/voltage change on anything over any hardware according to X sensors**. F@H drives it flat out normally... **Though I have to admit I've no idea what values they should be but 29C for a CPU/39C motherboard seems fine to me. As you say Linux is quite resilient but presumable if there is a problem it should be possible to get it logged somewhere?
Tom te tom te tom

I don't think I missed the point - perhaps I was a little vague though,
my apologies.

When you don't need the processing power, F@H will be utilising at least
one of your CPU cores to 100% (depending on how you set it up),
performing large scale iterative calculations - day to day computing
just doesn't use your CPU like this. If your CPU computes one wrong
value in a desktop program you may never know about it, although it
might crash eventually - in an iterative calculation it will invalidate
everything which follows.
And I'm not making myself clear.
Its a one core steam driven machine. With F@h running the machine always uses 100% cpu. So cpuburn/prime etc dont work the machine any harder.
Only f@h seems to go wrong:
So 1) I get a lot of duff fah runs..
or 2) there is a problem that doesnt seem to affect any other program so either the programs, or the kernel fix the problem and move on.

If its 2 then either f@h has a small problem (not likely) or I should be able to get the kernel to log the problem for me How do I do this sensibly - I dont want to know every step it takes just the bit where it does something to 'correct' an error - if indeed it has?
Tom te tom te tom

--
The Mailing List for the Devon & Cornwall LUG
http://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html