Errors - 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
23 Jun 2014 3:12:02 UTC
Topic 84968
(moderation:
I have received some errors lately, for app:
Gamma-ray pulsar search #3 v1.11 (FGRPopencl-nvidia)
The error is:
Outcome Computation error
Client state Compute error
Exit status 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
7.4.2
exceeded elapsed time limit 136.04 (300000.00G/2205.26G)
Is the app's rsc_fpops_bound value set incorrectly?
http://albertathome.org/workunit/604516
http://albertathome.org/workunit/604518
http://albertathome.org/workunit/604531
http://albertathome.org/workunit/604554
Errors - 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
)
We're in the middle of Boinc server software testing here, see the news threads, the rsc_fpops_bound is O.K, the server is supplying ridiculous speed estimates for the initial tasks.
Claggy
You need to edit
)
You need to edit client_state.xml to supply 100x higher rsc_fpops_bound values, to get you past the 11 validations needed to get APR to drive rsc_fpops_est.
Albert is running vanillia creditNew code and David is using GPU peak flops to estimate GPU speeds, which turned out to be a pretty daft assumption.
We'll be moving to a redesigned Credit Scheme over the next few weeks or so - you can try and keep up with the 'project code updted' news thread for that development.
Queen of Aliasses, wielder of the SETI rolling pin, Mistress of the red shoes, Guardian of the orange tree, Slayer of very small dragons.
Thanks guys. I just found
)
Thanks guys. I just found that thread, and subscribed to it. I just figured that an issue like this deserved to be reported in its own thread.
I do not plan on editing my client_state.xml file, because, if Albert fixes it server-side, I'd like the ability to test that fix.
Thanks again,
Jacob
Trouble is, if you don't take
)
Trouble is, if you don't take precautions like that, you'll never complete any tasks, and never be able to explore any other aspects of the server code.
I'm expecting that when the server code is next updated, I'll let it run for a few more hours to check we haven't introduced any new bugs, then force my 'graphing' host to get a new HostID and do it all over again with a completely clean application_details record.
I didn't want to spam the boards with my stats - just milestone theads - but apparently signatures are no longer optional. Follow the link if you're interested.
http://www.boincsynergy.com/images/stats/comb-3475.jpg
It's no trouble for me. It's
)
It's no trouble for me. It's just wasting a bit of my resources. I am not attached to work out any additional server/scheduler bugs. I am attached simply to test that units complete. And I had a problem with that. And I reported it.
Until server-side implements a fix, it looks like the units will continue to waste resources. It is unfortunate. I hope they fix it.
RE: It's no trouble for me.
)
Well, if you continue to waste resources after you've been told what's going on and why, be our guest. It's your electricity bill.
It's going to be fixed, though it may not be "they" that does it.
I didn't want to spam the boards with my stats - just milestone theads - but apparently signatures are no longer optional. Follow the link if you're interested.
http://www.boincsynergy.com/images/stats/comb-3475.jpg
I am happy to be your guest,
)
I am happy to be your guest, as you and the team figure out the scheduler issues.
I have the same errors, but
)
I have the same errors, but my wing(wo)man with nVidia cards also have this error. If done by a CPU then it is validated. So I think it has something to do with the GPU app.
Only Gamma-ray pulsar search #3 v1.11 (FGRPopencl-nvidia) have this error (at my side).
Greetings from,
TJ.
The error is due to an
)
The error is due to an initialisation bug in the creditNew code.
If you want to prevent it, you need to make a manual adjustment (edit) to client_state.xml. Specifically you need to increase the value for the GPU tasks by at least 2 magnitudes (add 2-3 zeros).
If you don;t feel confident enough for that, you can either try to juggle around with what tasks you get until the patch has been deployed, (e.g. opt out of GPU tasks), stop crunching for Albert until the patch is in or just let them error out [as Jacob is doing] - then when the patch is in it should correct itself.
In the first two, you need to monitor the news thread(s), for patch announcements.
Queen of Aliasses, wielder of the SETI rolling pin, Mistress of the red shoes, Guardian of the orange tree, Slayer of very small dragons.
Specifically: Fetch some
)
Specifically:
Fetch some FGRP (Gamma-ray pulsar search) work.
Exit BOINC completely
Edit as Eyrie describes. You'll find it in the definition for each of the tasks you've downloaded.
Restart BOINC, and allow the tasks to run and report as usual. Probably best to set 'No New Tasks' while you do this.
Once you've reported and validated 11 tasks, the procedure should no longer be necessary. If you didn't get 11 validations from the first batch, repeat as needed.
I didn't want to spam the boards with my stats - just milestone theads - but apparently signatures are no longer optional. Follow the link if you're interested.
http://www.boincsynergy.com/images/stats/comb-3475.jpg