| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Bionic Applicatrion selection algorythem? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 10
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
So I've set my profile to take any work that needs to be done (even beta) and run it as it's handed off to my system for processing.
How is work selected for being distributed and dispatched? Is it a FIFO queue or is there a priority queue ? Is there a set of queues that are for those machines with a certain capacity? Is there a queue for some members (who have proven their machines run all the time)? I'm just kind of curious. While I recognize all the work is important to someone's research, and working on the projects benefits us all (in the long run anyway) are there some research Bionic views as having higher priority that others? TIA ---Barney |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
No idea where folk get Bionic, but it is not a legal program afa WCG is concerned.
----------------------------------------WCG work distribution is not set in stone: 1. First the specialised 1 project crunchers get work 2. Then the Multi-project get a share of e.g. RICE+DDDT 3. What's left of each project goes into the pool and gets a pro-rata weight 4. Beta only goes to proven machines with short return times (low/no buffer). This work is seldom and separately announced in beta forum. The initial shared memory buffer is 1,800 work units in size pretty much handing out according the FCFS principle, so sometimes work for the fav.project is depleted at which times the "alternate work" option kicks in (if selected). The pool rarely has equal share work, as what is created by the work generator depends on what comes back completed from the clients. Then there are situations where one or the other batch is sped up as the scientists are waiting to complete experiment results. The Techs than use their joystick to push a particular project batch through the system. End of the day, the goal is for all projects to get their even share, but practically it's not possible, so this is how it looked yesterday July 10 in CPU years, which totaled 178. 36,05197 FA@H 38,24194 HPF2 0,00000 HDC 0,00000 GC 0,00000 HCMD 27,13517 DDDT 0,00000 AC@H 40,47793 HCC 37,07233 NRW Sort of the picture for those crunching with BOINC. ciao
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Jul 11, 2008 2:33:37 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks for the explanation... so the values you provided at the bottom.. were those returned work elements or work elements awaiting to be dispatched for processing?
Again, thanks for the explanation... A couple of days ago I had 2 cores running both saying "high priority" but I didn't quite know what that meant... and thus far I've not located anything that described how to understand the meaning associated when viewed from the WCG Bionic "Tasks" tab on the local machine. I've been kind of waiting to see that message again... and haven't thus far. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello BarneyBadass,
'High Priority' means that your BOINC scheduler thinks that the deadline is very close. If you see this and the deadline is far away, then something has probably happened to the values of the factors used by the scheduler to figure out how fast it can expect to run. The usual cause is that it thinks you only run your computer a small percentage of the clock time. Lawrence |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Hi again,
----------------------------------------We only get to see the total numbers for completed work, which in years was the above. The amount of work in the feeders is not public. It's enough to keep us going till at least 2013. High priority can mean 2 things: -You have jobs that are under threat due receipt of rush work -You have buffered too much, crunch too intermittent that the client comes to think that you wont be able to finish any job in time, which could even be the last job in the queue. Rule than is Earliest Deadline First order. There's loads of Wiki's around on BOINC and none of them actually is complete, rather they are somewhat complementary. This one googled on EDF: http://www.boinc-wiki.info/Earliest_Deadline_First . The scheduler and meaning is too complicated so best put on your reading glasses, search and learn. Start here: http://boinc.berkeley.edu/wiki/The_BOINC_Manager
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
thanks to both of you...
I was just curious about how work units were selected for dispatch ... It's interesting to watch all this stuff... the graphics don't do much for me unless I could understand the particular (_fill in the blank here_) I can't imagine how much cpu the graphics must consume that could be used by the cpu.... but I doubt it's insignificant say in one 168 hr contiguous block of time. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hmmmm....
so the next question is... about the longest running work unit I've seen has been about 7 hrs... can one specifically request the bigger units of work? Heck... I don't care if it takes a day or more to complete; my systems are up 24/7 anyway... so getting those work units that are larger and require longer run times wouldn't bother me in the least. TIA ---Barney (and just because I'm fat doesn't mean I'm purple!  ) |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Nutritious Rice for the World is currently the only project providing variable length work units.
Each NRW unit is sized to run for exactly 8 hours. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
interesting... the rice project seems to only take 7hrs on my system
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As you can see, it's rock steady. R00037_ 823f54c85933fa57b1d231e5c71b05dc_ 02_ 001_ 6-- Azrael Valid 07/02/2008 17:39:59 07/11/2008 20:39:01 8.01 56.7 / 54.5 R00037_ 9b40c48c9a3ca3dadc1dac440cec3dcc_ 00_ 003_ 13-- Azrael Valid 07/02/2008 17:38:47 07/10/2008 13:04:01 8.05 57.0 / 52.4 R00037_ 9b40c48c9a3ca3dadc1dac440cec3dcc_ 00_ 000_ 16-- Azrael Valid 07/02/2008 17:38:47 07/05/2008 17:47:42 8.04 55.1 / 54.5 R00037_ 9b40c48c9a3ca3dadc1dac440cec3dcc_ 01_ 000_ 15-- Azrael Valid 07/02/2008 17:38:47 07/05/2008 17:47:42 8.02 55.1 / 54.3 R00037_ 9b40c48c9a3ca3dadc1dac440cec3dcc_ 01_ 006_ 1-- Azrael Valid 07/02/2008 17:38:47 07/08/2008 06:24:56 8.04 56.6 / 52.8 |
||
|
|
|