I have been happily running BOINC since March including CPDN. Over the last couple of days I have been running some unrelated processes for my university degree - I am a doing a cryptography unit and these processes have been doing brute force attacks on some ciphertext. This has obviously been taking up most of my HTT CPU. However, I left BOINC running in the background assuming that it would be okay. All projects with the exception of CPDN are okay. CPDN has gone astray - three CPDN WUs have failed with an "Unrecoverable error for result WORK_UNIT ( - exit code -5 (0xfffffffb))" message.

I am taking a guess here - is there an assumption that a CPDN WU will complete some work after a period and if it doesn't it aborts? If so, that is a very bad assumption to make because CPDN will be a very low priority task and will hardly ever get any CPU time if there are cpu-intensive tasks utilising the CPU. Am I right - is there such a check? If so, can I suggest it gets fixed and uses the time actually spent on the CPU rather than including sleep time?

"I am taking a guess here - is there an assumption that a CPDN WU will complete some work after a period and if it doesn't it aborts? If so, that is a very bad assumption to make because CPDN will be a very low priority task and will hardly ever get any CPU time if there are cpu-intensive tasks utilising the CPU. Am I right - is there such a check? If so, can I suggest it gets fixed and uses the time actually spent on the CPU rather than including sleep time?"

This assumption is not correct.

Not sure what is wrong, but it may be some incompatibility between CPDN and your decryption task, or if there has been a change in the temperature of the room the computer is in so the PC is running hotter. Compared to the other BOINC projects, CPDN is more sensitive, being longer and more computationally intense.

Thanks for replying.

> This assumption is not correct.

Okay I am willing to accept that but could you please explain how the check works then?

> Not sure what is wrong, but it may be some incompatibility between CPDN and
> your decryption task,

I can't see how there could be an incompatibility between a java program (my crypto one) run by one user and a BOINC program run by another (I use a dedicated account to run BOINC). For all its faults, I don't think processes within Windows XP SP2 interfere with each other.

> or if there has been a change in the temperature of the
> room the computer is in so the PC is running hotter.

Not an issue. It is cold here at the moment. The room would be about 18C.

> Compared to the other
> BOINC projects, CPDN is more sensitive, being longer and more computationally
> intense.

Sensitive to what? I accept that the WUs are longer but I don't accept that they are more "intense". All BOINC tasks (at least that I have running) will use the full resources of a CPU if available.

All the other projects kept working fine - only CPDN WUs failed. That makes me think that CPDN is at fault.

Hi Peter

Exit code -5 means a calculation error of some sort. It is a known problem - the people in Oxford do know about it. When this happens, it's just as well that the model aborts. Geophi is right in saying that of all the boinc projects, cpdn is the most sensitive to any problems.

Did your machines previously complete whole cpdn models? If they did, then why not just back off cpdn while you're doing the university work, while continuing the other boinc projects? Then come back to cpdn later? Because..........some people have spent quite a lot of time getting round this -5 error. As in your case this is probably just a temporary problem, it wouldn't be worthwhile to spend time on it.
__________________________________________________

> Because..........some people have spent quite a lot of time getting round this
> -5 error. As in your case this is probably just a temporary problem, it
> wouldn't be worthwhile to spend time on it.

Hi. I am having the -5 error now that I'm using hadsm4.13. I even upgraded to BOINC 4.72, but every time, it crashes *instantly* after downloading the work unit. Even when the computer is doing nothing else. Even though I've been crunching for CPDN since way back into the beta stage and I've almost never had a crash, yet I've had four crashes in three days since it upgraded itself to 4.13. What to do? Your comment above leads me to think there is a fix. Any ideas?

Thanks.

Michael
<img border="0" src="http://www.boincstats.com/stats/banner.php?id=2922" />