| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Human Proteome Folding - Phase 2
Thread: It's taking soooo long! |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 19
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I've never seen workunits go this long on my computer before. Any idea what's going on?
Workunit Status Project Name: Human Proteome Folding - Phase 2 Created: 01/29/2008 13:41:59 Name: lm505_00002 m505_ 00002_ 10-- Valid 01/30/2008 18:48:28 02/04/2008 08:15:06 30.58 117.2 / 80.6 Usually, my claimed credit is very close to the granted credit. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello qt314159,
30.58 hours is unusually long. HPF2 occasionally gets caught in an endless loop that only ends when a reboot causes it to restart from the last check point (or until it times out with an error). This is a bug that we have never been able to track down. Perhaps that happened here. Lawrence |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Cant be that Lawrence. The restart/resume looses all the looping time and progress percentage from the last good checkpoint.
----------------------------------------Personally not had a hanging HPF2 job since i last moaned in the CA room. The science versions have not changed, so possibly its tougher work we had (the plots suggest so). Not all machines being equal this could be a case where the machine had a particular hard time. qt314159, can you please go to the Result Status page, find the work unit and click on the name. can you please post the quorum list so we can see if your run time was just exceptionally long compared to the others which could explain why your claim was 'normalized'. cheers
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Feb 4, 2008 1:08:50 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The endless loop bug mentioned by Lawrence is alive and well. I encountered another one a few days ago.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Mine was extraordinarily long, compared to the others. I'm not complaining about the amount of points. I understand how that works.
----------------------------------------The only reason I mentioned it was because, normally, my computer is right in the thick of the average - i.e., very close to the amount of points awarded. When there's a large discrepancy, I check to see my processing time. Here's the list: Result Name Status Sent Time Time Due / Return Time CPU Time (hours) Claimed/ Granted BOINC Credit lm505_ 00002_ 15-- Valid 01/30/2008 18:58:53 02/01/2008 02:14:19 14.17 94.8 / 80.6 lm505_ 00002_ 5-- Valid 01/30/2008 18:58:25 01/31/2008 07:44:34 5.07 67.9 / 80.6 lm505_ 00002_ 1-- Valid 01/30/2008 18:57:53 02/02/2008 10:32:01 7.14 71.7 / 80.6 lm505_ 00002_ 13-- Valid 01/30/2008 18:54:52 01/31/2008 07:15:21 7.22 81.7 / 80.6 lm505_ 00002_ 17-- Valid 01/30/2008 18:54:23 02/01/2008 13:57:44 4.84 89.9 / 80.6 lm505_ 00002_ 8-- Valid 01/30/2008 18:51:38 01/31/2008 07:10:58 5.58 62.9 / 80.6 lm505_ 00002_ 14-- Valid 01/30/2008 18:51:17 02/01/2008 05:21:35 8.31 82.4 / 80.6 lm505_ 00002_ 0-- Valid 01/30/2008 18:51:10 02/01/2008 12:38:33 11.44 89.0 / 80.6 lm505_ 00002_ 12-- Valid 01/30/2008 18:51:04 01/31/2008 09:22:12 7.26 75.4 / 80.6 lm505_ 00002_ 4-- Valid 01/30/2008 18:49:37 01/31/2008 06:50:30 3.20 93.2 / 80.6 lm505_ 00002_ 6-- Valid 01/30/2008 18:49:35 02/02/2008 15:26:31 8.03 82.8 / 80.6 lm505_ 00002_ 10-- Valid 01/30/2008 18:48:28 02/04/2008 08:15:06 30.58 117.2 / 80.6 <<== mine lm505_ 00002_ 2-- Valid 01/30/2008 18:48:08 02/02/2008 07:16:51 9.08 84.1 / 80.6 lm505_ 00002_ 18-- Valid 01/30/2008 18:46:55 01/31/2008 08:27:25 4.65 69.6 / 80.6 lm505_ 00002_ 16-- Valid 01/30/2008 18:44:14 02/02/2008 06:38:24 10.17 77.2 / 80.6 lm505_ 00002_ 9-- In Progress 01/30/2008 18:42:49 02/10/2008 18:42:49 0.00 0.0 / 0.0 lm505_ 00002_ 3-- In Progress 01/30/2008 18:22:55 02/10/2008 18:22:55 0.00 0.0 / 0.0 lm505_ 00002_ 11-- In Progress 01/30/2008 18:11:58 02/10/2008 18:11:58 0.00 0.0 / 0.0 lm505_ 00002_ 7-- Valid 01/30/2008 18:03:20 01/31/2008 20:10:39 5.14 69.1 / 80.6 I also have one that took over 30 hours and ended up being "inconclusive".  I've rebooted, and it didn't change the time on the ones that are processing. I'm thinking about avoiding the hpf project because of this. [Edit 1 times, last edit by Former Member at Feb 4, 2008 10:45:20 PM] |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7809 Status: Offline Project Badges: 
|
What are the specs on your system? And what are the last few results from hpf2 on your system. Just for comparison, I have 1.6ghz P4 which usually runs a hpf2 unit in 18 to 24 hours.
----------------------------------------Cheers
Sgt. Joe
----------------------------------------*Minnesota Crunchers* [Edit 1 times, last edit by Sgt.Joe at Feb 5, 2008 4:04:06 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
System spec/info: AMD Athlon 64 X2 3800+, 1GB RAM, but 256 assigned to video share, currently showing 510-520MB in use (reading 65MB boincmgr, 50mb for firefox, then dropping to under 20mb and down) and ~45MB swap file on a 2.2GB swap partition. running Ubuntu Linux 7.10 mostly as a dedicated cruncher, running 2 w/u simultaneously in boinc 5.10.8
Last few HPF2 have been very long, 21-30 hours (eg lo377_00008 - 29.74 hours - Valid), previous to that it was 12-16 hours (eg lo055_00027 - 14.86 hours, ln934_00026 - 13.30 hours). I just checked my upcoming w/u, and I have one with an estimate of over 43 hours (not started yet) with a deadline of 01/03/08, (lo452_00032_21) while I am currently 7 hours in / 8:45 to complete on lo555_00007_9 and 4 hours in / 11:02 to complete on lo562_00024_15, both with deadlines of 15/03/08. Looking at the completed results from other crunchers, they have times from 5 to 20 hours for this w/u, due to the short deadline and high estimated time, I'm very tempted to just abort it, and perhaps even deselect HPF2, if my system is not going to be good enough to crunch it in a reasonable time, especially considering this system is 99.9% crunch, doesn't get used for anything else usually. |
|||
|
||||
|
Cajun Cutie
Cruncher Joined: May 30, 2005 Post Count: 21 Status: Offline Project Badges: 
|
Yes, I too have experienced longer processing times in recent HPF2 WU's, but not as described above. Today I downloaded lo591_ 00000_ 19-- which had an original estimated time to complete of 27:40 hours. After it began to run, that estimate gradually reduced and it finally completed in 7.71 hrs. One other result is listed at 6.31; all others are still pending. So far nearly all recent HPF2 WU's have run on my machine between 7-9 hrs.
I have never experienced a hang or endless loop or incomplete. Errors have only resulted because I intentionally aborted a WU. I'm runing a Dell XPS 410 with an Intel 2.4GHz Core 2 Duo processor and 4Gb of ram. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Here is the thread announcing the change in HPF2 work unit lengths - http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=18589
Lawrence |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
It's a chain reaction. The first 'new' HPF2 had wrong flop counts in header (way too low) making them appear as short, but were needing triple the time of the estimate. Because of the mishap, BOINC adjusted the RDCF control (Result Duration Correction Factor) which is used to compute the time of all work coming after for WCG. When other work with correct flop count estimates started arriving after, BOINC applied the new, incorrect RDCF, causing for the that work to appear too long.
----------------------------------------It will take a day or 2 for that value to return to normal again depending how many work units are processed because RDCF goes up very fast, but goes down very slowly (BOINC cache safety feature to prevent over-buffering). Blame it on the Sun (Grateful Dead) 
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Feb 27, 2008 6:30:30 AM] |
||
|
|
|