| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Agent overestimating runtime |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 3
|
|
| Author |
|
|
david@madjbright.co.uk
Cruncher United Kingdom Joined: Jan 10, 2006 Post Count: 35 Status: Offline Project Badges: 
|
I am running Boinc on 7 systems. One of them is an old Win98 system which provides these messages when it starts up:
----------------------------------------Starting BOINC client version 5.8.15 for windows_intelx86 log flags: task, file_xfer, sched_ops Libraries: libcurl/7.16.0 OpenSSL/0.9.8a zlib/1.2.3 Data directory: C:\PROGRAM FILES\BOINC Processor: 1 GenuineIntel x86 Family 6 Model 8 Stepping 1 [x86 Family 6 Model 8 Stepping 1] [fpu sse mmx] Memory: 127.44 MB physical, 512.00 MB virtual Disk: 12.71 GB total, 5.36 GB free When it downloads a work unit it overestimates the time to complete by a huge margin. One unit had an original time to complete of over 800 hours while it completed in about 30 while the current unit is 50% through in 3:15 while the time to complete still reads 28:30. The problem is that with the Connect to network about every: set to 2 days this system does not download another unit so start and sits idle for some time between units. I do not really want to set the Connect to network about every to greater than 2 as all systems are permanently connected to the Internet when they are on, but are not on all the time and I do not want them downloading units that I will not complete in the required time. Any suggestions?
David
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It will "learn". Give it a few more days, and if it's still off by lots, we'll help you adjust it manually.
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi,
----------------------------------------we've seen these odd cases of overestimation before. BOINC has a complex scheduling and if a job e.g. is computed to do 30 hours and once or twice you only computed a few hours in a day, it extrapolates in an accelerated manner. The effective progress is stored into a value called "Duration Correction Factor" (DCF). Let the machine run for a little while, a week or 2 and things will normalise. Alternately, go to the project tab and select to not fetch work. Let it finish the current job and visit the project tab again and do a project reset. That should set things right. After, set the option to fetch work again in the project tab. Let us know if things persist.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at May 3, 2007 6:06:36 PM] |
||
|
|
|