climateprediction.net home page
Very long deadline dates?

Very long deadline dates?

Questions and Answers : Getting started : Very long deadline dates?
Message board moderation

To post messages, you must log in.

AuthorMessage
Shawn85206
Avatar

Send message
Joined: 15 Mar 16
Posts: 2
Credit: 129,524
RAC: 0
Message 57419 - Posted: 1 Dec 2017, 14:57:21 UTC

I participate in eighth different BOINC projects. I was looking at the deadline dates for the four Climate Prediction units sitting on my laptop...mid November...2018???

Now I'll admit that I have not sat and watched, but overall my laptop seems to give work units with the closest due dates priority. Are these data sets really going to sit here until, the end of the first week of NEXT November? Is there any compelling reason that the due dates are so far out in the future?

Again, my understanding of how BOINC prioritizes data sets may be flawed, so am anxious for some explanation.
ID: 57419 · Report as offensive     Reply Quote
Profile Iain Inglis
Volunteer moderator

Send message
Joined: 16 Jan 10
Posts: 1079
Credit: 6,904,049
RAC: 6,657
Message 57420 - Posted: 1 Dec 2017, 15:42:34 UTC

The CPDN work units are relatively long by BOINC standards and the deadlines are longer because of that. However, the deadlines do seem seem to be set unrealistically far in the future - so far away that I doubt a model returned at that time would be of any use to the researchers, who will get most models back quite quickly. The problem with shortening the deadlines is always said to be that CPDN then starts to exclude other projects on a PC that is running more than one BOINC project. I’ve never investigated BOINC deeply enough to know how much the deadlines could be reduced before that would happen.

Welcome to the CPDN message board!
ID: 57420 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 57422 - Posted: 1 Dec 2017, 17:08:53 UTC

Building on Iain's reply: In CPDN's early association with boinc (first non-Berkeley project in boinc) there were complaints aboout CPDN dominaing the resource. Hence, adjustments were made and a fictitious deadline.

Results are desired ASAP = the real deadline.
"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 57422 · Report as offensive     Reply Quote
Profile JIM

Send message
Joined: 31 Dec 07
Posts: 1152
Credit: 22,053,321
RAC: 4,417
Message 57423 - Posted: 1 Dec 2017, 19:15:37 UTC - in response to Message 57422.  

One reason the deadlines are so long is the models used to be much longer. Back in the old days (10+ years ago) models ran from 1920 to 2060. They could take 8 or 9 months to process (running 24/7) on a 1.2 GHz processor. Back then teams used to report when one of their members finished a model.

The only way I know of to get Boinc to share between CPDN and other projects is to set the work buffer very short at less than 2 days. Then it will ignore the deadlines and run the work in the order it was downloaded and allow the percentage of time it spends on each project to be governed by the “Resource” settings in the “projects” tab of the Boinc manager.
ID: 57423 · Report as offensive     Reply Quote
Profile Alan K

Send message
Joined: 22 Feb 06
Posts: 484
Credit: 29,579,234
RAC: 4,572
Message 57424 - Posted: 1 Dec 2017, 23:57:47 UTC - in response to Message 57423.  
Last modified: 1 Dec 2017, 23:59:25 UTC

It's also a good idea to tweak the proportion of time BOINC assigns to each project. For example I run CPDN, Milky Way and WCG in the ratio 150:40:10 which effectively gives CPDN 75% of the work time. This has to be done in each project's computing preferences on your account home page, not in BOINC.
ID: 57424 · Report as offensive     Reply Quote

Questions and Answers : Getting started : Very long deadline dates?

©2024 climateprediction.net