I'd definitely go with 100% throttle for a wee while to see if it goes away. That would help the techs to identify a lingering issue. Could be specific batch specific. Has happened in past.

Second thing is to not use the science screensaver or the graphics display. Again it would help to identify, assuming all your display drivers are up to date. There's a post in Start Here, Information about BOINC on the display items to check: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=4925#75214

Check the computer power supply.

It will resume at the last checkpoint.

We have not discussed the actual Windows and Security software. What are they? Also in my signature is a 'Issue?' checklist. You may want to go thru and see if there is anything you think might be a co-reason.

Why you loose all the work instead of the resume at last checkpoint is that you set the writing too frequent ( 2 times a minute). Try setting it to 10 minutes (600 seconds). If writing happens while computer crashes, chances are the science file gets damaged. Checkpoints are usually 10 minutes or more apart so the writing every 30 seconds does not help with WCG jobs.

You understood correct. By extending the write period to now 900 seconds, the chance of crash happening simultaneous with a disk save is reduced. Once we know the cause you can go back to the default writing frequency (which i personally do not like as disk i/o takes higher priority cycles, which can impair the user experience and increasingly, the slower the machine is).

wanted to edit my post:

i switched FAAH to ubuntu (with the beryl xgl thing) computer
wich is more powerfull barton 2.1 gig
with 1gig ddr .
and left proteins@home on the pentium III 1gighz 1gig memory
(the one with reboot problems).


each computer will run on his own project now.

and third computer amd x2 works on both sometimes.

see ya