Einstein task will not succeed
30 Nov 2007 22:47:03 UTC
Topic 193356
(moderation:
I have an Einstein task running on this PIII computer, 512megs DRAM, 1 gig CPU, against my better judgment, because people said it would be fine, that BOINC would "figure out" what this computer could do. That was several weeks ago. BOINC has not figured out anything.
The task has 20 hours left to run and must finish by 2:57PM 12.1.07. That is just not going to happen.
I run four other projects on this computer. Obviously BOINC sees this project as most "deserving". I don't. Tasks are supposed to run 60 minutes and then turn over to another project. No other project is getting any run time.
Unless someone can tell me how to keep Einstein playing by my rules, it will be gone.
Einstein task will not succeed
)
When any project's task is in deadline trouble, BOINC will only allow that task to run. Since your deadline isn't over yet, just let it run. The task will be re-issued to another computer, but if you deliver it before that one does, you will still get the credit for it. (If you're into that, the data returned is always of use)
RE: RE: Tasks are
)
With all due respect, my issue is that it is my computer, I want it to be used the way I choose, so I guess that leaves me out of Einstein.
If Einstein can not give me tasks for this computer that can live comfortably with tasks from other projects where the time is much less of a burden, then I was correct several weeks ago when I said that I should remove Einstein from this machine.
So, can anyone tell me how to get tasks from Einstein that are in the competance range of a PIII, 512megs DRAM, 1 gig CPU?
http://sciencesprings.wordpress.com
/http://Facebook.com/sciencesprings
http://twitter.com/sciencesprings
RE: So, can anyone tell me
)
All einstein tasks are in the competance range of a 1ghz PIII.
With that computer it is unlikely that you will be able to mix more than 3 projects without running into deadline problems. BOINC will be able to accommodate any number of projects. HOWEVER there will be times when there is not work from all of the projects on your computer or one project runs continuously to meet it's deadline. In these cases the client will pay back the time to the other projects by not downloading more work from the project that had a deadline issue. Your resource shares will be honored over a longer period.
Reducing your queue to a bare minimum may help. However I do not think you will be able to achieve what you desire, the more projects you have the more likely you will not have work from some of them at any given time. For example the host I am typing on is attached to 34 projects and currently has work from 10 of them on hand, with no deadline pressure.
BOINC WIKI
BOINCing since 2002/12/8
RE: With all due respect,
)
As Keck Komputers mentioned, your system could complete on time, however there is a "gotcha". When you attached there was a resource share / allocation that you could assign. Did you do that? If not, it was assigned the default of 100, which goes into the pot of all the other resource shares that you have. BOINC is figuring it out, which is why it has now shifted into "EDF" (although not strictly called that anymore). Once it completes the task, it probably will know that your resource allocation for Einstein is too low, so you may not even be able to get another task, much similar to what happened to me with an LHC work request:
2007-10-31 18:11:36 [lhcathome] Sending scheduler request: To fetch work
2007-10-31 18:11:36 [lhcathome] Requesting 2282 seconds of new work
2007-10-31 18:11:41 [lhcathome] Scheduler RPC succeeded [server version 505]
2007-10-31 18:11:41 [lhcathome] Message from server: No work sent
2007-10-31 18:11:41 [lhcathome] Message from server: (won't finish in time) Computer on 98.1% of time, BOINC on 100.0% of that, this project gets 1.0% of that
2007-10-31 18:11:41 [lhcathome] Deferring communication for 1 days 21 hr 1 min 10 sec
2007-10-31 18:11:41 [lhcathome] Reason: requested by project
It's up to you, but since your system is that close to meeting deadline, you can probably "fix" the issue by giving Einstein another 1-3% resource allocation.
@Bernd, in case you are reading - It would be a good idea to get SSE optimization into the Windows app soon so that issues like these are less likely to happen. If not, then may I suggest going ahead and putting the deadline at 3 weeks again until such time as the optimization is included? If not, then how about a deadline of 18 days instead of 14?