| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: OpenZika
Thread: super low credit for some super long Zika WU |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 44
|
|
| Author |
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
The stats shows these 'assemblies' are in small minority as the Zika graph hardly has the daily average move. What's different in these computed ligands? Credit average is still 'excessively' high at 32.97 for Sunday, but maybe the techs did knob something in the crunchers' favor with the newest app version ;?)
----------------------------------------[Edit 1 times, last edit by SekeRob* at Jul 11, 2016 9:57:41 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The stats shows these 'assemblies' are in small minority as the Zika graph hardly has the daily average move. What's different in these computed ligands? I agree with the small minority - I rarely catch a long one. Comparing the Result Log with others from the same batch downloaded at the same time shows no obvious difference apart from the long one having 190 tasks versus 15-25 tasks for those running around 2 hours.I think we're just part of a big computer game run by Keith  |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
The game ideally being having an objective to sent those heavies to more powerful brethren, but the feeder not yet having the skill to make out the M6 from Silverstone CPUs ;D
----------------------------------------(Is the machine not learning anymore, or are those ligands off the chart?) [Edit 1 times, last edit by SekeRob* at Jul 11, 2016 11:23:41 AM] |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3716 Status: Offline Project Badges: 
|
Things are "improving" : a new monster is currently running on my tthrottled ultrabook, 338 tasks (!!) estimated at 20,118 GFlops while it should reasonably be more than 300,000!!!
----------------------------------------Another crediting frustration in view when its estimated 2.5-3 days elapsed will finish... Sigh... Edit: Adjusted the expected total elapsed time. ---------------------------------------- [Edit 1 times, last edit by JmBoullier at Jul 12, 2016 11:23:47 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
And I'm helping to keep the average length down with this one:
ZIKA_ 000037049_ x5jmt_ ZIKV_ NS3helicase_ B_ 0433_ 0-- Microsoft Windows 10 Professional x64 Edition, (10.00.10586.00) 708 Valid 11/07/16 07:13:13 12/07/16 13:22:24 0.03 1.6 / 1.6 Number of tasks = 1 |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2346 Status: Offline Project Badges:
|
Lowering the average length some more ...
ZIKA_000037114_x5jmt_ZIKV_NS3helicase_B_0084_0-- Linux 4.2.5-300.fc23.x86_64 708 Valid 11/07/16 09:52:28 11/07/16 12:05:37 0.02 1.5 / 1.3 [13:09:55] Number of tasks = 1 [13:09:55] Running task 0,CPU time at start of task 0 was 0.000000 [13:09:55] ./ZINC02294016.pdbqt size = 25 2 ../../projects/www.worldcommunitygrid.org/zika.x5jmt_ZIKV_NS3helicase_B.pdbqt size = 4288 0 [13:11:33] Finished task #0 cpu time used 77.883559 |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges: 
|
Identical problem yesterday with WU ZIKA_000038141_x2jlv_DV4_NS3_ssRNAdelMn_chA_B_0395:
----------------------------------------48 points for 18.87 CPU hours on a i7 4770K at 100% (no throttle). Usually this host earns 35 - 40 points hourly per core. Yves |
||
|
|
SekeRob
Master Cruncher Joined: Jan 7, 2013 Post Count: 2741 Status: Offline |
Totally befuddled why we have this extreme [do-not-add-up] task packages with just 1 through IIRC a max of 350. What's the decision process that causes this to happen [end of a compound/parameter test set?] Machine learning still not captured all ligand variations? If unknown ligand, why not take the low road and package few, so the run-times don't go off the scales and not cause outlier penalty on the hard crunches?
Anyway, with the new app and credit stats having to rebuild at client level, still see a high 7 day project mean of 30.94/Hr and a last day of 28.66/Hr... well above the WCG overall averages, so it's still a credit party [for most :]. (graphs through my sig link... when I manage to upload fresh copies again :). |
||
|
|
Rickjb
Veteran Cruncher Australia Joined: Sep 17, 2006 Post Count: 666 Status: Offline Project Badges:
|
I'm only crunching Zika on Linux-x64.
Program file is wcgrid_zika_7.08_x86_64_pc_linux_gnu. Just had a look through 616 valid results. Nil long WUs and credit looks fine. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The biggest I've crunched yet:
ZIKA_ 000038961_ x2jlx_ DV4_ NS3hlcs_ RNAdelMnChA_ 0392_ 0-- Microsoft Windows 10 Professional x64 Edition, (10.00.10586.00) 708 Valid 14/07/16 14:18:11 16/07/16 22:07:26 25.14 48.0 / 48.0 "Only" 236 tasks. Even the Result Log was truncated at the start (details of task #0 and part of task #1 missing). And look at the joke credit. Compare with my next result on the same machine: ZIKA_ 000039450_ x2jly_ DV4_ NS3hlcs_ ssRNA_ chA_ A_ 0350_ 0-- Microsoft Windows 10 Professional x64 Edition, (10.00.10586.00) 708 Valid 15/07/16 11:40:17 16/07/16 22:09:51 1.41 48.2 / 48.2 |
||
|
|
|