How are your temperatures? Don't forget that this makes your CPU run at 100% and therefore you need to make sure they don't run too hot. Generally make sure that your heatsink is on properly, you're not blocking any air vents, keep insides clear of dust, and ideally have some decent case fans.

I have experienced different temperatures on different projects on a different grid initiative which will remain nameless... but I don't know the reason why? Initially it made some boxes crash and I moved them to only run the other one.

That would be my first guess anyway. If you post again with your hardware details make, model, CPU, OS, cooling, etc then someone may know of a more obvious answer?

Failing this, go into your device manager and device profiles and there you can set all your devices to only run Rosetta until the problem is solved. You can then continue to run the project.

just a reminder that some computers need MORE than 256 MB RAM to run FA@H. one of mine did, because part of the RAM was already for graphics or something. it did run proteomes. when it got a FA@H problem it would pause the agent very quickly.

The difference between FAAH and HPF is RAM, not temperature. FAAH can use high memory that is ordinarily left alone. Several people have discovered previously unsuspected memory errors when running WCG applications. Looking at the 'Useful Utilities' thread at http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=2490 I suggest:
1) Motherboard Monitor: http://mbm.livewiredev.com/
Monitors CPU temperature. If the manufacturer does not supply a temperature monitoring program for your board, then see if this program works with your computer. If the temperature is too high, then read http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=2683

2) CPU Tester: http://7byte.com/index.php?page=download
General diagnostic program

3) Memory Tester: http://www.memtest86.com/
Memory diagnostic program

Added: Virtual Memory needs another 300 MB for FAAH. Check in your control panel on at least one problem computer to make sure that the default is enabled to expand VM if required. This is probably not your problem.
mycrofth

Not sure how much VM is needed in total but I checked one machine that has 1GB RAM and it has 1.5GB VM. Not much is happening on that box other than running this agent.

I do not have time to check them all right now. I have all device profiles set up to not get any FAAH wu. When time permits, I will create new device profiles and transition servers one at a time to see what happens.

Please use the support email system and send me the device ID numbers (from the preferences window of the agent) for the machines having problems and I will see if there is anything in the logs that might give any clues. Thanks.