Advanced search

Message boards : Number crunching : Not Happy

Author Message
Gentilli
Send message
Joined: 8 Jan 09
Posts: 12
Credit: 15,378,064
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 16250 - Posted: 12 Apr 2010 | 17:00:18 UTC

Hi, I am not happy that I wasted 9+ hours on a WU that I consider as defective (WU ID 1351175). It has gone bad on two computers. And I have to loose credits for this...............
____________

[BAT] Annabel
Send message
Joined: 27 Aug 08
Posts: 5
Credit: 4,307,023
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 16258 - Posted: 12 Apr 2010 | 19:30:49 UTC - in response to Message 16250.

Well ...
There is only one solution for not having bad WU's at all and that is ... never crunching and not using computers.

Gentilli
Send message
Joined: 8 Jan 09
Posts: 12
Credit: 15,378,064
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 16262 - Posted: 13 Apr 2010 | 0:56:20 UTC - in response to Message 16258.

Unfortunately I think that that day might be closer at hand then I think.

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 16271 - Posted: 13 Apr 2010 | 8:40:25 UTC - in response to Message 16262.

Unfortunately I think that that day might be closer at hand then I think.


You make a valuable contribution to this project don't let a couple of bad results stop you we all get them from time to time for a myriad of reasons.


____________
Radio Caroline, the world's most famous offshore pirate radio station.
Great music since April 1964. Support Radio Caroline Team -
Radio Caroline

Gentilli
Send message
Joined: 8 Jan 09
Posts: 12
Credit: 15,378,064
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 16299 - Posted: 14 Apr 2010 | 14:47:18 UTC - in response to Message 16271.

Thank you for the encouraging words.

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 16350 - Posted: 17 Apr 2010 | 8:29:42 UTC - in response to Message 16262.

Unfortunately I think that that day might be closer at hand then I think.

As others have said, there is an inherent risk to running any task for any project. The risks increase if the project and or task is part of a beta run and the rise even higher if an Alpha test is being run. In the case of GPU applications the driver version and the BOINC version have impacts also...

So, when we run projects that have multi-hour or multi-day tasks our risks increase that bad things can happen and the consequence is the loss of credit earnings.

This is one reason I have long been an advocate of running as many projects as possible in part because there are so many that need our help. Moreover, you get a much broader mix of tasks with most taking less time than is "normal" here ...

On the system I am looking at right now, I am running one GPU Grid task that looks to be 10 some hours in length ... not sure why, if it is the task, the new Win7 installation, the fact that I am running Nvidia and ATI cards in the same system, or just a long task ... if it fails, well, 10 some hours shot until I can figure out the issues ... I have a similar problem with the 5870 ATI card and MW tasks not validating (until I stopped MW work on that system, and since I have not seen an all clear yet, I am still not doing MW work on my fastest system because the failure rate was too high).

In the mean time, on that same system, I am pumping out Collatz tasks with about 6 minutes per task on the ATI GPU ... If I start MW again those will run in even less time ... in the mean time the CPU cores are hard at work on the other 50+ projects I am attached to (just added a couple tonight, though one has no work today)...

Remember, the point is the science, and science has failures ... credit is only our way to measure how much valid science we have done ... but in no way reflects any virtue... what I mean, is that the fact that I have done "x" amount is really quite meaningless in the grand scheme of things ...

So, the point, you have not seen the last failure ... there will be more ... how many and when is the only question ... I hope you stick with it ... just remember if the failure rates are higher than you like ... try another project ... there are quite a few out there and I even just found a new one that supports both GPU classes (DNETC@Home) though I am not sure how well they run (yet) ...

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 16351 - Posted: 17 Apr 2010 | 8:41:21 UTC - in response to Message 16350.

It is a cyclic theme that results from project development and card aging.
I use to run my GTX260sp216 on Folding@Home, but out of the blue it stopped working. I tried different computers and different operating systems, but it still would not work. It works perfectly here!

I had a couple of cards that use to work on GPUGrid but stopped, so I put them on MilkyWay. MW changed their requirements to CC1.3, so I had to put them on folding@home for a while.

In the end I sold them and used the return to buy cards that work well here.

Post to thread

Message boards : Number crunching : Not Happy

//