Advanced search

Message boards : Multicore CPUs : redundant result??

Author Message
sam
Send message
Joined: 30 Apr 09
Posts: 15
Credit: 228,425
RAC: 0
Level

Scientific publications
wat
Message 12362 - Posted: 4 Sep 2009 | 20:35:54 UTC

http://www.gpugrid.net/workunit.php?wuid=736231

what is that? the WU on client-side crunches for a while and then presents, aborted by project. no credit. what the heck is that especially since on the serve-side it presents redundant result???

the way it looks: the WU goes out more than once and the first one to get it done gets the credit, the other is screwed. I hope I am wrong with that!! and I also hope this is more about questionable reporting than the way it appears.

provided for documentation purposes only.

Toni
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 9 Dec 08
Posts: 554
Credit: 4,273,184
RAC: 0
Level
Ala
Scientific publications
watwatwatwat
Message 12382 - Posted: 5 Sep 2009 | 16:19:59 UTC - in response to Message 12362.

Please see http://www.gpugrid.net/forum_thread.php?id=996

sam
Send message
Joined: 30 Apr 09
Posts: 15
Credit: 228,425
RAC: 0
Level

Scientific publications
wat
Message 12446 - Posted: 10 Sep 2009 | 13:46:53 UTC - in response to Message 12382.

"As a general rule, workunits are NEVER canceled when they are running so that you don't lose any credits or compute power.

WUs may be canceled only while they are sitting in your queue, waiting to be executed. This is normal, and in fact it is a feature of the BOINC scheduler which prevents "redundant results" to be computed."


I have had three WUs with no credit--a noteworthy number in less than a month when crunching 6-7 or week. one for supposedly being late, no credit yet the results uploaded, and two redundant results--significant time crunched on both and yet no credit. I do not get how your explanation applies to the ps3 that is permitted only 2 WUs. if I have a ps3 with 2 WUs and they way ahead of the deadline, your answer does not suffice. are you suggesting that boinc is aware of the WUs sent to everyone else? if so, why was I sent WUs that would be killed as redundant and why crunch, for hours? again, way ahead of the deadline. if there is threat of redundancy wasting many hours crunching, why send WUs in such a way to risk that?

also, there are irregularities with credit granted. please look at the following. http://www.gpugrid.net/workunit.php?wuid=754122 the WU payload is the same size as every other that I have seen on this ps3, took the same time, yet, granted far less than claimed and this not the first time. do you have an explanation for that or a link?

thanks.

Jari Pyyluoma
Send message
Joined: 2 Aug 08
Posts: 12
Credit: 1,147,651,344
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 12467 - Posted: 12 Sep 2009 | 6:34:49 UTC - in response to Message 12446.

I do not know if this is the same phenomena - I had one machines RAC get stuck at about 3000, while another yielded about 5000. This went on for some (longer) time, and I was wondering why this was. However, now both machines are very close to each other, at about 5000. I do not know if it was the fact that I reinstalled Boinc, or just that I restarted the weaker one, that made it a better producer. Never got around to take a deeper look at things, just got this feeling that maybe some of the SPEs had crashed or something. I could not see any pattern to this anomaly when just looking at the credits. At one point I got this ridiculous idea that it was some server-side reward system for machines that had worked diligently for a longer time, because it was the newer machine that was underachieving.

Post to thread

Message boards : Multicore CPUs : redundant result??