| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: The Clean Energy Project - Phase 2 Forum
Thread: Unfair points for completing work unit |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 12
|
|
| Author |
|
|
ca05065
Senior Cruncher Joined: Dec 4, 2007 Post Count: 325 Status: Offline Project Badges: 
|
E214414_ 838_ C.36.C30H16N2O2SSi.02119284.1.set1d06_ 1-- 640 Valid 7/12/13 10:41:25 7/13/13 08:11:43 9.05 266.5 / 139.4
E214414_ 838_ C.36.C30H16N2O2SSi.02119284.1.set1d06_ 0-- 640 Valid 7/12/13 10:28:19 7/12/13 10:31:55 0.05 1.9 / 21.5 Mine is the _1 unit which ran for over 9 hours while my wingman only ran for 3 minutes. My requested points were nearly halved while my wingman's were credited eleven fold. His unit completed in job 1: <core_client_version>7.0.28</core_client_version> <![CDATA[ <stderr_txt> INFO: No state to restore. Start from the beginning. [04:28:24] Number of jobs = 16 [04:28:24] Starting job 0,CPU time has been restored to 0.000000. [04:30:59] Finished Job #0 [04:30:59] Starting job 1,CPU time has been restored to 151.593750. Application exited with RC = 0x1 [04:31:27] Finished Job #1 [04:31:27] Starting job 2,CPU time has been restored to 177.734375. [04:31:27] Skipping Job #2 [04:31:27] Starting job 3,CPU time has been restored to 177.734375. [04:31:27] Skipping Job #3 . . . How can such early completion be considered valid? |
||
|
|
armstrdj
Former World Community Grid Tech Joined: Oct 21, 2004 Post Count: 695 Status: Offline Project Badges: 
|
The sub jobs in a Clean Energy workunit can have dependencies on the output from other sub jobs. When a subjob fails the code will continue on and run any other subjobs that are not dependant on the output of the failed sub job. Your wingman failed on sub job 1 and therefore could not run job 2 which is typically where the most CPU time is used. We are working on a code upgrade for the application used by the Clean Energy Project and will look for a solution to address this issue.
Thanks, armstrdj |
||
|
|
Werinbert
Advanced Cruncher United States Joined: Jun 13, 2013 Post Count: 56 Status: Offline Project Badges: 
|
Then it begs the question of what is a valid result? ca05065 returned some work for which there is not a wingman to validate (the current wingman skipped). Is it just simply ignored by WCG and ca05065 gets credit only for the parts the wingman didn't skip? Wouldn't it be better to just mark the wingman as invalid and get a new wingman to run the task thus validating the entire WU?
---------------------------------------- |
||
|
|
armstrdj
Former World Community Grid Tech Joined: Oct 21, 2004 Post Count: 695 Status: Offline Project Badges:
|
CEP2 does not run redundant workunits except occassionaly to verify hosts. For CEP2 we use an alternative method to validate results outside of the scientific application. On the researchers side they are able to cluster results and determine if any results are invalid. In the cases where workunits are run redundantly the researchers use the most complete result.
Thanks, armstrdj |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Don't want to open another thread, so I'll ask here: why is the max CP time I get for any WU is 12 hours??
 Thanx in advance. Regards, Igor |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Don't want to open another thread, so I'll ask here: why is the max CP time I get for any WU is 12 hours?? It's a preset cutoff time for all clients running CEP2. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Besides all that, the credit system is driven by a One Armed Bandit and for each science a different bandit is programmed to most always give you 3 rotten cherries as outcome... Figure that, momentarily all my CEP2 jobs get 2-3 times as much credit as FAHV, which gives a pathetic 10-11/hr, even when these run in 64 bits, which CEP2 can't. This system is so broken at the root, and compounded by those OAB rules per science, that the word 'system' really is best stricken from the WCG dictionary, a complete and utter waste of time to question or reflect on.... over the years a grand source of irritation to which sledge and wooden stick is best applied.
--//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Tony, Rob, thanks for the answer. I guess I'll just cry a little and calm down
 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Tony, Rob, thanks for the answer. I guess I'll just cry a little and calm down Don't forget that you will receive full points and run-time credit for that 12 hours of CPU time. However, you could also investigate why the elapsed time, typically 17 hours, is so much larger than the CPU time of 12 hours. That's only 70% efficiency. If you were achieving say 98%+ efficiency, then your WU throughput, points and run-time increase would be as good as that machine can give you on CEP2. Of course, if you're deliberately throttling the CPU below 100% to reduce temperatures, then you may not want to improve the situation. |
||
|
|
cjslman
Master Cruncher Mexico Joined: Nov 23, 2004 Post Count: 2082 Status: Offline Project Badges: 
|
hhhmmm... interesting topic. I'm not much of a CEP2 cruncher, but I'm trying to get a little in, here and there. So far, the few CEP2 WUs I have crunched have been well behaved and usually finish in a 10-12 hour time frame, but now I have one WU which already has 10 hours crunched and is showing another 15 hour to go
---------------------------------------- . Is this a normal CEP behavior or is there something special that I should know about?CJSL Crunching for a better world... |
||
|
|
|