Long running work unit

[philip-in-hongkong]

I have a long running WU 6330886 seems stuck at 98.765%. It once runned to nearly 100% with no running time left but now jumped back to 2 days 3 hours remaining run time and increase slowly at about 1 second for every 1 minute run. This WU has gathered 175d CPU time and it is sad to abort it now. Any advice on the action taken? Indeed I can let it run if there is worth to do so. Lastly, if I abort the WU, will I get any credits?

[Jacob Klein]

If it is still using CPU, and BOINC still has increasing "Elapsed" time, then my advice is to KEEP RUNNING IT (DO NOT ABORT).

For these tasks:
- They were developed to go up to ~100%, then show "98.765%" constantly until completion.
- "Remaining (estimated)" time is not relevant.

Some examples of tasks I've completed are:
- 98.765 at 85.0d; completed successfully at 204.1d
- 98.765 at 109.1d; completed successfully at 262.8d
- 98.765 at 218.2d; completed successfully at 302.7d
- 98.765 at 383.9d; completed successfully at 544.1d
- 98.765 at 144.6d; completed successfully at 226.9d
- 98.765 at 163.6d; completed successfully at 353.3d

I have a few others, too, that are at 98.765 and still being worked on, including a WHOPPER:
- 98.765 at 192.2d; currently at 545.2d

So ... your task may have a lot of work to go still, not sure.

[philip-in-hongkong]

Thank a lot. I will let the WU run till it finishes.

[ChristianB]

Yes, please do send them to me -- I am highly interested!
If you teach me how to do the job, I may volunteer to manually compare all that need manually compared!

This is more a problem of scientific accuracy. From a technical point I can already provide the textual diffs of both result files (VM and non-VM) that are not considered valid. Some of them seem simple but some differ in the scientific output. The hard part is to check if the scientific differences are still within the desired error range and can be considered valid.

I'll send you a PM with the last set of results for cross-validation.

[Jacob Klein]

Yes, please do send them to me -- I am highly interested!
If you teach me how to do the job, I may volunteer to manually compare all that need manually compared!

This is more a problem of scientific accuracy. From a technical point I can already provide the textual diffs of both result files (VM and non-VM) that are not considered valid. Some of them seem simple but some differ in the scientific output. The hard part is to check if the scientific differences are still within the desired error range and can be considered valid.

I'll send you a PM with the last set of results for cross-validation.

I received your PM, and have already finished processing the results. I sent you a PM reply.
Basically, it took me 2.5 hours to manually compare 24 results, so ... if you have more, I volunteer!
Really looking forward to clearing out some of these work units that are waiting on manual validation!

Thanks,
Jacob

[Jacob Klein]

RacerX conquered another one - Completed and validated!
http://www.rnaworld.de/rnaworld/workuni ... id=6330897
178.9d

He's now actively looking for another one, to replace it.

[ChristianB]

I've gone through the remaining XXL/VM cross validation tasks and cleared all but one. The one VM task had an empty result archive but there is still a task in progress so this will show up in the next round. This is now kind of automated and I just need to execute some scripts and look at some textfiles. I'll try to find the hour each month to do this on a regular basis.

[Michael H.W. Weber]

Excellent!

Michael.

[Jacob Klein]

I have just completed my longest-running-task!

Zathras (an old laptop), successfully completed this MONSTER:
http://www.rnaworld.de/rnaworld/result. ... d=14949860

CPU Time:
555.9 days

Later today, he gets his reward -- an upgrade of the OS and newer VirtualBox

[Michael H.W. Weber]

Very nice.

Michael.

[gemini8]

Congrats again, Mr. Monster-Hunter!

[Jacob Klein]

Speed, my newest PC, just completed his largest task:
http://www.rnaworld.de/rnaworld/workuni ... id=6330936
276.1d