| 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: 50
|
|
| Author |
|
|
cio_redulla
Advanced Cruncher Philippines Joined: Apr 24, 2006 Post Count: 130 Status: Offline Project Badges:
|
Hello,
----------------------------------------I would like to know how to request for additional work units. Sekerob told me at the BOINC Forums that crunchers who have demonstrated the ability to crunch more [than the maximum of 10 WUs per request] can request larger allotment. I've set the "Connect to server value" top 5.5 days. I would like to download more than 10 WUs per day so that the BOINC agent in my computer won't have to connect to server often. The 5.5-day value is not being followed because the agent has to connect once the 10 WUs are finished. Example: 1. Download 10 WUs [max per request per 10 minutes] 2. Download another 10 WUs after 10 minutes, until total number of WUs for the day is equal to 30 to 50. After obtaining 30-50 WUs, don't connect to server until 5 days has passed. I don't know why the BOINC agent won't download more WUs even though 10 minutes had already passed after the last 10 WUs were downloaded. The number of WUs in the Task panel of the BOINC manager only stays between 10 to 12 WUs and never more than that. ![]() |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
To be precise, up to 120, u get 10 at the time. If u have set a large buffer like your proposed 5.5 days, BOINC should automatically take care of fetching u more work, but it will send them in sets of 10. What it does after the first 10 is compute how much time your particular machine would take to complete the work according the benchmark and a few other 'performance' criteria that BOINC maintains locally. If it than determines that your buffer is not filled up, BOINC would request more.
----------------------------------------120 is a protection rule. To get more in a day u would have to send a mail. why you are not getting more in your buffer than 10-12 WU's has very much to do with the actual work. If after letting BOINC alone for a few days, nothing is improving while staying permanently connected, it could be that some values have somehow gotten inflated like the Duration Correction Factor (DCF). Than it's suggested to set the option 'do not fetch new work', work off the tasks in the tasks buffer and than do a Reset in the projects tab. How many seconds of work is being requested? You should see that in the message tab right before a download starts. If it's a very low value or a highly inflated value like millions of seconds, the mention reset is the path to follow. Oh, and i do hope you dont have created a general_prefs_override.xml file. U can change anything in the web profile, but if a work buffer value is in that file, it will supersede. U'd have to change the override file and force BOINC to read the prefs again thru the Update option. long winded, but hope there is an answer somewhere in here. PS1 Normally if you buffer 5.5 days, u are not, by logic expected to be connected for that period. U can though schedule a daily connect of e.g. 1 hour..... that should be enough to send back results and fetch new work to back fill the buffer. PS2 The 5.8 version has a manual 5 minute override feature. Even if u told BOINC to 'connect per schedule', if u hit update, it will do its upload/download thing for 5 minutes or for as long as upload/download is ongoing (i think, but never observed). Same goes for disconnecting. It will stay an extra 5 minutes connected. cheers
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jan 12, 2007 10:46:17 AM] |
||
|
|
cio_redulla
Advanced Cruncher Philippines Joined: Apr 24, 2006 Post Count: 130 Status: Offline Project Badges:
|
Thanks for the info Sek!
----------------------------------------I'm waiting for the agent to finish crunching the remaining WUs. I'll reset the project and I'll be observing everything. By the way, if I hit the Update button in the Projects tab I only get a "not requesting new work or reporting completed tasks" message. ![]() |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Cio, that message is standard ..... BOINC has a mind of it's own. Often 10 minutes later it starts pulling work, if it sees room in the buffer. Some of the design is to make it the least burdensome on the servers and there are multiple, that perform different tasks in a BOINC environment.
----------------------------------------cheers
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I'm having some problems with one of my machine running BOINC 5.8.2 not pulling down enough work units. I have set my profile to request 2.5 days of work (<work_buf_min_days>2.5</work_buf_min_days> in my global prefs file) and recently allowed all the work units to complete and report and then reset. The machine requested 21600 units of work "1/13/2007 4:34:27 PM|World Community Grid|Requesting 216000 seconds of new work", but it only pulled down work estimated to complete in 13 hours (1 genome unit and two FAAH units). My DCF is now 1.00 (before reset it was around 1.15). I've checked my local override file and it does not have an ovveride on number of days of work. Any suggestions on how to fix this?
|
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges:
|
What operating system? It would make a difference in the BOINC version we're talking about.
----------------------------------------I tried running either an alpha or beta 5.7 BOINC version on Windows and could never get more than ten workunits at a time. They would run completely down to nothing before sending 10 more, regardless of the number of days set in the network field. I reverted to the production 5.4.11 BOINC version and the queue happily filled up to where I had set it. When one or a few workunits complete, more work comes in when I am connected.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 1 times, last edit by retsof at Jan 13, 2007 11:03:51 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Windows XP. In this one machine (work profile), it waits to get more work until all the units complete. Running 5.8.2 (with Windows XP) on other machines using different profiles, it gets more work units after a unit or two complete.
|
||
|
|
keithhenry
Ace Cruncher Senile old farts of the world ....uh.....uh..... nevermind Joined: Nov 18, 2004 Post Count: 18667 Status: Offline Project Badges:
|
Do the profiles all have the same "Connect to server every X days" value? It sounds like the Work profile has the 0.1 default value.
---------------------------------------- |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
3 Things to note:
----------------------------------------1. The Flop estimates were adjusted on the various projects a few days ago. 2. Work like the HDC have are actually taking much longer presently on my 2 machines that do it. 3. 5.8.2 is beta (in feature lock down) but tweaking has not stopped. The flops are used to compute the TC (To Complete Time). 5.8.1 was fine, but 5.8.2 coupled with the adjusted estimates has caused the 2 BOINC installs' DCF (Duration Correction Factor) to yo yo up and down like would u believe. Because i upgraded right when the Flop estimates were changed, cant say if the cause is the latest version. I'll just sit back, but must say that after 2 days the trickle has resumed and the sum of work buffer time estimates more or less equals my work buffer setting. The DCF is now at 1.34 where in past it was always hovering around 0.8 on the P4 HT. The C2D is momentarily 1.6. The higher the value, the less work is buffered. In relative terms the latter actually is computing the FAAH's in stock at 10/1.6 from the TC, so as said work trickle is resuming. Oh, and when the WCG technicians do their tweaking, they do that in consideration of the supported 5.4.9/5.4.11 version!. Too much brain cringing for a Sunday morning and just seeing that the 06:06 UTC stats run did not happen it does ![]()
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jan 14, 2007 2:37:15 PM] |
||
|
|
cio_redulla
Advanced Cruncher Philippines Joined: Apr 24, 2006 Post Count: 130 Status: Offline Project Badges:
|
What operating system? It would make a difference in the BOINC version we're talking about. I tried running either an alpha or beta 5.7 BOINC version on Windows and could never get more than ten workunits at a time. They would run completely down to nothing before sending 10 more, regardless of the number of days set in the network field. I reverted to the production 5.4.11 BOINC version and the queue happily filled up to where I had set it. When one or a few workunits complete, more work comes in when I am connected. I have the same situation as retsof's. I had already finished resetting the project and it's still the same-I couldn't download anything more than ten WUs at a time. The computer has to finish crunching the ten WUs before downloading another batch of 10 WUs. I'm running BOINC 5.8.2 for Windows and my DCF is 0.926484. I guess Sekerob and retsof are both right, the particular version of BOINC running in the computer has something to do with it. Should I go back to 5.4.11? ![]() |
||
|
|
|