• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Issues with BOINC

LocutusX

Diamond Member
Hello, I installed BOINC about 2 weeks ago and have been running SETI on it off and on for a while. I'd say I have about 4-5 days of dedicated computer time on it now. Since my rig is an A64 I'm sure I have a fair number of wu's completed. Now what I want to know is why I'm still at 0.0 credits? Did I misconfigure something in boinc?

Also if I'm not mistaken it takes longer to complete a wu in BOINC as compared to SETI classic?
 
Seti for BOINC was comparable speed-wise to v3.08 "classic", meaning roughtly 10% slower than v3.03. But the new v4.05 (win-only) is for some reason even slower. Most probable reason is some debug-code was left on, or code-changes...
This really doesn't mean much, since the crediting is based on crunch-time, and even with the slower v4.05 BOINC is more efficient than "classic" since only sends the same wu to 3 users normally and not 7+ as "classic" is doing.


As for no credit, the validator has been back-logged for a long time, and seems to have been turned off the last couple of days meaning no credit to anyone. Also, there's no credit before 2 other users also has crunched the same wu, so even if everything works it can take upto 14 days before all has returned the result, even longer time if someone reset/detach/miss deadline.
Also, uploading results is a 2-step-process, there the result-file is normally immediately uploaded then finished crunching, but the servers doesn't know of this before the result is Reported. Results is reported next time you're connecting to the scheduling-server, either due to asking for more work, nearly reached deadline, or if manually "update project".

The preference "Connect to network about every x days" will normally cache between x & 2x days of work, so because of the 14-day-deadline don't set x higher than 7. Also, since in v4.05 the wu takes longer than expected to crunch, I'll recommend to not set this higher than 5 for the moment.

Any upload-errors or other problems can also be the reason for no crediting, this will show up under Messages-tab in the gui, or stdout.* sterr.* in BOINC-directory, but unfortunately service-install haven't currently any logging.

A last reason can be, if you're overclocking, you're overclocked too much so you're only returning garbage, and BOINC doesn't give credit for this garbage...

edit - If you're using BoincView or another similar monitoring-tool, you'll get the message-info even if installed as service. 😎 More info on boincview, here
 
Thanks for all the info, and:

Originally posted by: Rattledagger
A last reason can be, if you're overclocking, you're overclocked too much so you're only returning garbage, and BOINC doesn't give credit for this garbage...

I use an 8 hour run of both memtest86+ and Prime95 torture test to "validate" my overclocks. I assume that's good enough for ensuring a reliable BOINCing machine?

In fact, I've done my "burn-in" test about 3 or 4 times, on separate occasions, at my current overclock. I hope it's good enough.

 
Well, one user at BBR posted a couple days ago one "stable" overclocked machine having no problems with 24h-memtest86 & 12-24-hour prime95 but failed miserably on all LHC-wu... Running default speed fixed the crashes, showing it really wasn't stable even it passed these burn-in-tests...
http://www.broadbandreports.co...ark,11544122~mode=flat

Seti v4.05 is according to some running hotter than previous versions, so it's possible this will push a previous system over the top. The BOINC-client is reportedly also crunching a little differently through a wu but should get the same end-result, so even stable in "classic" doesn't mean you're not getting problems in the BOINC-version. Actually, since in classic you've getting credit for all wu really the only way to know pushed the overclock too high is if starts to crash all wu after a couple of minutes...


CPDN is also known for pushing many "stable" overclocked systems over the edge, so even passing memtest86 & prime95 isn't a guarantee shifting to another project will not lead to only returning garbage.

Of course, since currently in seti@home it's impossible to know if a result passed validation or not, it doesn't need to be overclocking-errors but one of the other reasons you've still not gotten any credit.
 
Back
Top