| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 12
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
BOINC continuously adjusts it's buffering behavior to the actual progress made. Take a break of say 12 hours in the middle of a job and it will include that aggressively into account. It will take days of continuous crunching to return to - what the user considers - 'normal' up/download patterns. A typical value to check in the client_state.xml file is the Duration Correction Factor (DCF). If it's for instance 2.134563 it means that the device is deemed to take 2.1 times as long in finishing a job, than the actual estimated computational effort would be. There's a few more values that can hint at jobs taking on wall-clock longer than in CPU time, but wont bore you with those.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The way it works is rather clever. It will request enough work to just fill 3 days (according to its estimates, of course). Since work units come in discrete chunks, the actual amount of work sent will be more than 3 days. As soon as the buffer falls below 3 days again, BOINC will try to get an Internet connection to top it up*. However, if none is available, then it can continue running for exactly 3 days before actually running dry.
So, if you connect at least every 3 days, you should be fine. * If you have a permanent connection, BOINC will request 1 second of work at this instant and in return will get a single work unit. Dialup doesn't tend to work like that. |
||
|
|
|