| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: BOINC To Completion madness |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 33
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Distributed Computing (project).
The DCF is used in the buffer size calculation. If it is inaccurate, BOINC may estimate you need more or less work than can be done in the time period. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello schepers,
I am probably the least knowledgeable about BOINC, but I think that BOINC is dealing with 2 different servers. It sends results back to the project server and then sends a message to the BOINC scheduling server. It seems to wait a while before sending a message to the BOINC scheduler to tell it that it has sent a result back to the project server. I need to read a complete description of just how BOINC is supposed to work. But it may be very misleading to believe the 'ready to report' status. Lawrence |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Spot on, Lawrence.
schepers, the sad truth is some of the BOINC preferences are considerably more than they seem. Deciphering them is some kind of dark magic. An accurate description would be something like: "Setting this setting thus does (a), (b) and (c), except once in a blue moon when it will do (d). Unless is is lower than (f) or conditions (g) and (h) prevail, when it will do (i) or (j). However, if (k) or (l), then the setting will have no effect." Obviously, I've simplified a little.... Or, in summary: "min work buffer" and/or "connect to servers every..." are both insufficient explanations. |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges: 
|
Respective the DCF discussion of KH'.... the number is adjusted continuously and u need to let your machine alone, truly alone, for it does take weeks for BOINC to find it's balance. Whilst WCG has only one DCF for all it's projects and there for having an inherent issue, each DC has it's own DCF. e.g. my Tanpaku DCF is 1.05, the WCG is 0.8, the SIMAP is 0.85 and Rosetta 1.01.....guess who's doing the best flop estimations ;>) Yep, leaving things alone is normally best. The numbers are dropping slowly - VERY slowly. It just got so out of whack so fast that it got me wondering about this again. Each time I do that, it seems I manage to pick up another tidbit or two on the method to the madness behind it. I have been wondering if the fact that my "queue" got refilled after the outage in a pretty tight timeframe versus the more typical one or two out, one in over time. I was down to crunching on my last WU when the servers came back. Having reset the project (I believe I've only done that once before), had to wonder if that impacted it. I need to pay closer attention during odd events like that  |
||
|
|
Alther
Former World Community Grid Tech United States of America Joined: Sep 30, 2004 Post Count: 414 Status: Offline Project Badges: 
|
An accurate description would be something like: "Setting this setting thus does (a), (b) and (c), except once in a blue moon when it will do (d). Unless is is lower than (f) or conditions (g) and (h) prevail, when it will do (i) or (j). However, if (k) or (l), then the setting will have no effect." Obviously, I've simplified a little.... I think you've just described the operating instructions for a past girlfriend of mine. Clearly those were over simplified... Joking aside, I couldn't have explained BOINC settings better myself. I'm rolling here.  I think we just found the person we need to write our updated documentation. 
Rick Alther
Former World Community Grid Developer |
|||
|
||||
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges:
|
Interesting.....my To Completion numbers have dropped about 20 percent overnight (within the previous 10-12 hours). What's different? I don't have any HDC WU's at the moment.
---------------------------------------- |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Usually the TC follows the time of the last completed project e.g. the last FA@H + a drag factor. If u happened to have a very quick FA@H in between the number drops rapidly for all the FA@H in stock, but increasing goes faster as the work buffer algorithm is conservative to ensure that what is cached, does get finished within the connection time.... i.e.. the exampled 1.0 days.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
BobCat13
Senior Cruncher Joined: Oct 29, 2005 Post Count: 295 Status: Offline Project Badges: 
|
each DC has it's own DCF. e.g. my Tanpaku DCF is 1.05, the WCG is 0.8, the SIMAP is 0.85 and Rosetta 1.01.....guess who's doing the best flop estimations ;>) I recently ran an alpha project with very poor flop estimations. It caused my DCF to be over 20000, which made the to Completion estimates about 6 years instead of the actual 3 hours. |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges:
|
Usually the TC follows the time of the last completed project e.g. the last FA@H + a drag factor. If u happened to have a very quick FA@H in between the number drops rapidly for all the FA@H in stock, but increasing goes faster as the work buffer algorithm is conservative to ensure that what is cached, does get finished within the connection time.... i.e.. the exampled 1.0 days. Right. I believe the last WU I completed overnight was the last HDC in my queue. I'll have to try to remember to watch what happens the next time I get sent an HDC WU. edit: ah....just checked.....my DCF has dropped from 2.06 to 1.58 too edit 2: fcg's have processed now, down to only FAAH, TC looks to have dropped by the same margin again, DCF down to 1.28 now....will definitely have to check again when I get another non-FAAH WU and see how things change.....could the multi-project factor account for this much difference?? ---------------------------------------- [Edit 2 times, last edit by keithhenry at Jan 4, 2007 5:41:07 AM] |
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges:
|
Well, my TC times are down to something pretty realistic now. DCF is down to 0.98 as well. Clearly, TC's will go up much faster than they will come back down. Still, I am not sure if something I did kicked them up into space or if it was something else. Maybe one of these days, I'll actually understand the madness
---------------------------------------- |
||
|
|
|