| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Nutritious Rice for the World
Thread: What does "Inconclusive" mean for a Rice WU? [Answered] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges: 
|
AFAIK, "inconclusive" is awarded for other projects' WUs when the results from the initial 2 crunchers of the WU disagree, and applies temporarily until a result is returned from a 3rd or subsequent cruncher, matching 1 of the original 2 results. In this scenario, all WUs sent out are the same.
----------------------------------------With NRW, I thought that each cruncher received a different part of the total WU, but I don't know how results are checked. Here are the first 12 results listed for my first NRW "inconclusive", mine being last: Result Name - - - - - - - - - - Status - - - - Sent Time - - - - Time Due - - Return Time - - CPU Time (hours) - - - Claimed/ Granted BOINC Credit R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 6-- In Progress 07/06/2008 09:01:26 07/18/2008 09:01:26 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 9-- In Progress 07/06/2008 02:20:01 07/18/2008 02:20:01 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 8-- In Progress 07/05/2008 20:37:11 07/17/2008 20:37:11 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 16-- In Progress 07/05/2008 20:34:42 07/17/2008 20:34:42 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 2-- Inconclusive 07/05/2008 20:07:10 07/06/2008 12:09:55 8.01 161.2 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 4-- Inconclusive 07/05/2008 20:06:10 07/06/2008 10:19:16 8.00 146.3 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 10-- Inconclusive 07/05/2008 19:54:58 07/06/2008 13:01:15 8.00 160.1 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 18-- Inconclusive 07/05/2008 19:51:38 07/06/2008 09:16:39 8.02 131.3 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 14-- In Progress 07/05/2008 19:51:17 07/17/2008 19:51:17 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 5-- In Progress 07/05/2008 19:51:08 07/17/2008 19:51:08 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 12-- In Progress 07/05/2008 19:50:34 07/17/2008 19:50:34 0.00 0.0 / 0.0 R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 0-- Inconclusive 07/05/2008 19:50:04 07/06/2008 10:48:44 8.02 41.6 / 0.0 I thought that the pieces may have been sent out & crunched sequentially, in duplicate streams, but the Sent Times suggest that the pieces get crunched in parallel. ...? [Edit 2 times, last edit by Rickjb at Jul 7, 2008 6:47:42 AM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Your analysis is correct. Each of the initial 19 are unique but do carry certain verification algorithms to ensure that those not meeting the QS are rejected. Which one or more that is/are cannot be determined until any combination of 10 completed results sufficiently agree. Look in your result log and see if any of the lines is of the benign type, there's a Start Here forum FAQ listing the most common, or actually of an alarming nature.
----------------------------------------It's also possible this particular job does not compute and all turn into inconclusive. Then the system will withdraw the task set, send a note to the technicians who then will see if there is a special condition not covered in the application. [Added: 12 inconclusive does not bode well. You can see from the submission times (sort by Sent time) if extra copies were distributed.... the top 2 seem to be those] ttyl
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Jul 6, 2008 4:52:11 PM] |
||
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges:
|
Thanks, "Ace crusher from Earth without Winter (or inferno Summer)*" aka Sekerob ...
----------------------------------------WU status now Valid. The problem result, not mine, was: R00040_ 09fb04fcf8900fc71e5e6c4f1ab3d426_ 02_ 002_ 15-- Invalid 07/05/2008 14:50:06 07/05/2008 14:51:49 0.00 0.0 / 0.1 Agreement has apparently been reached, just as for "normal" projects. Next question: When all WU pieces are crunched simultaneously, how is duplication of work avoided? How do fast machines avoid overlapping with the work allocated to slower machines? How do slow machines avoid leaving gaps in the work by not completing enough steps? Or is the project probability-based (aka Monte-Carlo Method), where random possible cofigurations of the proteins are evaluated, and the best configuration is chosen at the end? *Google's translation to English of Sek's "Location" - I think he means cruncher, not crusher, and I assume he's concerned about climate change ... www.carbonequity.info [Edit 3 times, last edit by Rickjb at Jul 7, 2008 1:40:15 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Rickjb,
I am told that it takes from 60,000 to 600,000 structures predicted from one protein to adequately cover the solution space. Rosetta does this by using a different 'seed number' for each work unit. I suspect that NRW uses a similar technique. There are quadrillions of possible protein configurations that differ very slightly in the spacing of their atoms. Lawrence |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
That's a technical question which similarly is of importance for HPF2 (Rosetta) and obviously do not want i.e. duplication of effort. Only X out of millions is needed and each of the 19 tasks in the quorum are kicked out with a different departure point. The algorithm subsequently generates the seeds that wont be computed in any of the other 18.
----------------------------------------Sort of the simplified version. A new feature was added to the next BOINC that allows to set a number of structures needed. I've tested that version and in anticipation added the message generated in the log to the SH forum FAQ. Effectively when the required number is reached, e.g. 120,000, will the application finish off, thus not running 8 hours, but sooner. Not informed enough to know how e.g. the credit system will work, but figure that a 120,000 cap should give an equal grant to all (for those clients that manage to reach the max (or min. if you will). What it will do is speed up the project progress. [offtopic]Google does a reasonable job, but the literal translation did not quite get it right (and missed a part). Never use it to write to foreign officials ;-)[/offtopic]
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 3 times, last edit by Sekerob at Jul 7, 2008 9:01:45 AM] |
||
|
|
|