| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: UD Windows Agent Support [Read Only]
Thread: BOINC deadlines |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 6
|
|
| Author |
|
|
bobsmith
Cruncher Joined: May 6, 2007 Post Count: 9 Status: Offline Project Badges: 
|
Now that UD is being phased out, will BOINC deadlines be increased? The percentage of people who have their PCs on for enough hours in 10 days to complete a job must be small.
Bob |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
A thought that had crossed my mind, but think not and if at all not by much. When BOINC started at WCG it used to be 14 days and was sometime early last year cut to 7. Several projects have 9 days and HPF2 has 11. The problem is storage and moving thru the batches back to the scientists. With UD one just got the points and was done. With BOINC it means validation would have very long delays.... don't think many are waiting on that and with the ever faster computers, 9 days really means 1 hour a day average which is not a lot.
----------------------------------------Added: Statistics showed quite recently > 90% coming back within 4 days and 96% coming back within 6 or 7 days, so a cut off a 9 or 11 would let few fall thru.... the data volumes are just getting too big as WCG is growing while nearing it's 3rd anniversary.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Oct 8, 2007 6:42:05 PM] |
||
|
|
bobsmith
Cruncher Joined: May 6, 2007 Post Count: 9 Status: Offline Project Badges:
|
I have 3 PCs that do not get an hour a day, but still have decent processors. Also, if you shut down/logout rather than hibernating, a job may never reach a save point on an occasional use computer.
The second job that gets queued up has the deadline ticking away while it is waiting... I don't think the turnaround time is a problem for the research - I would think they need to wait until the end of the project to get the results. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
New work will not get retrieved unless BOINC calculates that the present job can be completed in time. Set the buffer (connect time) to e.g. 0.001 days, additional buffer to 0.0000 and new work would only be fetched when the present job is near ready. Also BOINC learns from actual crunching behaviour, so if just started, let it do it's thing and it will wise up.
----------------------------------------As for logging out interrupting computation, you are the prime candidate for the BOINC service install option. It would crunch whenever the machine is on. Added: Checkpoints on some of the jobs i run are quite short. Here a sample: 8-10-2007 20:05:19|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:10:50|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:17:13|World Community Grid|[checkpoint_debug] result li205_00071_17 checkpointed 8-10-2007 20:17:18|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:25:23|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:32:06|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:34:11|World Community Grid|[checkpoint_debug] result li205_00071_17 checkpointed 8-10-2007 20:39:16|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:47:03|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:53:17|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 20:53:35|World Community Grid|[checkpoint_debug] result li205_00071_17 checkpointed 8-10-2007 21:00:45|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 21:07:17|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed 8-10-2007 21:13:20|World Community Grid|[checkpoint_debug] result li205_00071_17 checkpointed 8-10-2007 21:14:27|World Community Grid|[checkpoint_debug] result li204_00024_12 checkpointed While papers may not be released until project completions + a year, there's a lot more on-the-go learning than we crunchers get to see. The projects have subdivision in many batches. If you visit e.g. the FA@H progress page, you see that many experiments have been run. The AC@H project is in fact so that the next batch cant be compiled until the current batch is completed.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 4 times, last edit by Sekerob at Oct 9, 2007 8:53:48 AM] |
||
|
|
bobsmith
Cruncher Joined: May 6, 2007 Post Count: 9 Status: Offline Project Badges:
|
Sounds good - how do I set it up like that? In the device profile web page I can only see "Connect to network about every", and no "buffer" or "additional buffer".
And what is a service install? Bob |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The service option is available during installation.
The additional buffer setting is new in BOINC 5.10, so when you upgrade you will both get this capability and be able to chose the service option. |
||
|
|
|