Page 2 of 2

Re: Project: 2665 (Run 0, Clone 910, Gen 7)

Posted: Wed Jun 04, 2008 1:41 pm
by Flathead74
Also from the FahWiki:

http://fahwiki.net/index.php?title=EUE_ ... orrections
Quit 101 - Fatal error: Too many box elem corrections

Quit 101 - Fatal error: Too many box elem corrections 1
...snip...
Folding@home Core Shutdown: EARLY_UNIT_END

This error is reported very infrequently, and with the limited reports received to date appears to be specific to certain WUs.

Re: Project: 2665 (Run 0, Clone 910, Gen 7)

Posted: Wed Jun 04, 2008 7:19 pm
by bruce
joce wrote:[16:32:33] stateQuit 101 - Too many box elem corrections 1
There's no doubt that this sort of an error MIGHT be caused by hardware, but it also might not.

On the science side, the temperature of the protein being simulated is, in fact, energy which causes random motions of all the atoms. In order for a protein to move from an unfolded state toward a folded state, it has to go through a lot of random motions, too. The systematic progress that the researchers are looking for is generally a challenge to identify because of all the randomness that is superimposed on the motion that they're looking for.

When the simulation starts, they must ASSUME the initial random velocities (with the help of some other software), but the sum of all those motions adds up to zero (if the protein is standing still) and to some other value if it's moving. If it's moving at a high velocity, the simulation must do box corrections so that the protein stays the center of attention rather than departing the area of interest. So in simple terms, the error message says that the protein is moving too fast.

Why is the protein moving too fast? Perhaps some unexpectedly strange motion would actually happen. Perhaps the assumed velocites at time=0 were "wrong" in the sense that they would never happen in real-life but they're close enough to a real-world possiblity that it's difficult to know. Perhaps a calculation error caused unrealistic velocities to be calculated. Perhaps . . . .

The calculation error is the most probable, but it's not certain. That's why we always ask if you just had one EUE or if you've had a number of them. Calculation errors tend to repeat, once some hardware problem causes them. The other possible causes for the EUE would generally be a single WU.

Re: Project: 2665 (Run 0, Clone 910, Gen 7)

Posted: Wed Jun 04, 2008 11:23 pm
by joce
EUE happens only 2 times for me, and the two times it was at the very beginning of the simulation, with the same WU :

Project: 2665 (Run 0, Clone 383, Gen 8)

Re: Project: 2665 (Run 0, Clone 910, Gen 7)

Posted: Wed Jun 04, 2008 11:27 pm
by bruce
joce wrote:EUE happens only 2 times for me, and the two times it was at the very beginning of the simulation, with the same WU
I'd assume that it's a bad WU and forget about it.