sample of log when work unit is sent back early

[veryhumid]

well looks like I finally did it, work unit got sent back early. Its really too bad, I just got my overclock stable after changing my GAT settings and was going to try for more. verbosity 9.

[14:21:17] Completed 103750 out of 125000 steps (83)
[14:27:49] - Autosending finished units...
[14:27:49] Trying to send all finished work units
[14:27:49] + No unsent completed units remaining.
[14:27:49] - Autosend completed
[14:30:29] Writing local files
[14:30:29] Completed 105000 out of 125000 steps (84)
[14:38:20] Quit 101 - Fatal error:
[14:38:20] Step 105965, time 211.93 (ps) LINCS WARNING
[14:38:20] relative constraint deviation after LINCS:
[14:38:20] max 0.000000 (between atoms 1 and 2) rms 1.#QNAN0
[14:38:20]
[14:38:20] Simulation instability has been encountered. The run has entered a
[14:38:20] state from which no further progress can be made.
[14:38:20] If you often see other project units terminating early like this
[14:38:20] too, you may wish to check the stability of your computer (issues
[14:38:20] such as high temperature, overclocking, etc.).
[14:38:20] Going to send back what have done.
[14:38:20] logfile size: 20157
[14:38:20] - Writing 20834 bytes of core data to disk...
[14:38:20] ... Done.
[14:38:20]
[14:38:20] Folding@home Core Shutdown: EARLY_UNIT_END
[14:38:24] CoreStatus = 72 (114)
[14:38:24] Sending work to server

I'm not sure if I get points or not, didn't have EMIII open.

 

[SewerBeing]

if its sending work, you are gonna get points, just not as much as finishing it would have given.

 

[veryhumid]

it would be nice if it was a proportion to what percentage I completed, that was a 90 pointer I think. Hopefully this just won't happen again...

 

[veryhumid]

crap just got another one!!! The only thing I have changed was my GAT settings to A-A-A-D-D from A-A-A-A-A, where my overclock made the whole computer unstable. Ahhh the life of a tweaker... WHAT A PAIN IN MY FOLDING ***!


LOL good profile quotes, It is definitely a fine day indeed for team 32

 

[Andyman902042]

Why don't you check for stability with prime 95 first?

 

[KrisMCool]

Both CPU overclocks and memory settings can cause errors like that (I've unfortunately seen them too when tweaking my rigs for best performance). Since GAT settings are what you changed last, that's what I would suspect (sounds like you already know that though). Tweaking GAT settings won't do anything to improve your folding performance. Prime95 is a great program to test for stability, but it can't catch everything. I've had instances where my OC tested stable in every bench and 24+ hr Prime, but would still produce errors when folding. It's too bad someone hasn't developed a program to specifically test for folding stability. Good luck finding your sweet spot.

 

[actong]

It's too bad someone hasn't developed a program to specifically test for folding stability.

What about stresscpu? (Scroll down to the 7th item.) You can download a precompiled windows binary at the bottom of this page.

 

[KrisMCool]

Thanks for the links. I'll give that a shot.

 

[veryhumid]

I'll let you know if i figure out exactly what eas causing the problem, i think it was definitely the memory settings. I think folding@home itself is a good stress testing application. It runs for long periods of time, uses 100% cpu and uses SSE instructions. I don't mind if I get a few work units sent back along the way to getting a good clock. I'll check out stress cpu, though, looks quite nice.