climateprediction.net home page
Posts by old_user502172

Posts by old_user502172

1) Message boards : Number crunching : Some LTD related CPDN questions (Message 32725)
Posted 23 Feb 2008 by old_user502172
Post:

It\'s basically a debug file, and won\'t exist unless you create it.



Thanks.

Jim
2) Message boards : Number crunching : Some LTD related CPDN questions (Message 32718)
Posted 23 Feb 2008 by old_user502172
Post:
Are those units in minutes?

Seconds.
Where is this file and is it a text file?

Sorry Jim. I assumed from your knowledge about LTD and the scheduler that you\'d also know about cc_config.xml. My mistake. The link Les provided tells you everything you need to know.


No worries, I\'m more comfortable with theory than with actually DOING something ;-)

I looked in my BOINC data folder and there is nothing there. Should there be? In fact, there is no cc_config.xml anywhere on my computer.

I\'ll have to see what program in Mac OS can serve as a plain text editor.

Jim
3) Message boards : Number crunching : Some LTD related CPDN questions (Message 32704)
Posted 22 Feb 2008 by old_user502172
Post:
If you set <rr_simulation>1</rr_simulation> in cc_config.xml you\'ll get a load of [rr_sim] messages telling you how long tasks are projected to take to complete (but not how long it is until they reach their deadline). The calculations are made every minute so you\'ll probably want to set the value back to 0 after you\'ve got one set of debug output (especially with 4 CPUs and that number of projects).

The line you\'re particularly looking for is something like
21/02/2008 17:43:10|climateprediction.net|[rr_sim] result hadcm3istd_0306_1920_160_05923447_5 finishes after 6302668.254245 (5279647.999301/0.837685)

Your system only has to go into high priority mode if you also get something like this
21/02/2008 17:43:10|climateprediction.net|[rr_sim] result hadcm3istd_0306_1920_160_05923447_5 misses deadline by 45.000000



Are those units in minutes? Where is this file and is it a text file? I am on Mac OS 10.4.11.

Thanks,
Jim
4) Message boards : Number crunching : Some LTD related CPDN questions (Message 32675)
Posted 19 Feb 2008 by old_user502172
Post:
I am running BOINC 5.10.34 for the Mac.


It almost certainly will go into \'high priority\' mode, given the 2% priority share. I don\'t know now long it\'ll take to realise that there is a problem (different versions of the Boinc manager behave differently).

5) Message boards : Number crunching : Some LTD related CPDN questions (Message 32668)
Posted 19 Feb 2008 by old_user502172
Post:
I am new to CPDN and have a couple of questions regarding the scheduler\'s handling of CPDN.

I have a 4 cpu Intel Mac. 50% of it is allocated to SETI, 25% is allocated to Einstein and about a dozen projects, including CPDN, share the remaining 25%.

When I attached to CPDN, two WUs were downloaded. One is due in about 4 months, the other is due in an eye popping 31 months. It appears that since there is no previous history with CPDN on my Mac, it has not accumulated any debt and therefore is running ahead all of the other programs that are sharing that 25%. Does that sound correct?

SInce I am allocating all of a princely 2% of the resources to CPDN, it seems natural to me that BOINC will panic early and go into what we now call high priority mode for the two CPDN WUs. My computer is on about 50% of the time which gives CPDN about 15 minutes/day/WU. If my WU is estimated to take 280 hours of CPU time, it will take 1120 days to finish the task. I only have about 120 days until the due date. How come it is not going into high priority mode? My guess is that since it still has very little debt accumulated, it still has priority so it doesn\'t have to go into high priority mode. Is that a correct assumption?

I had also assumed that LTD was recalculated every 60 minutes, which is my application switching time, or when a task ended. Is that a correct assumption?

This is not whinemail. I am happy to munch away on my projects and let BOINC figure out what should be running. This is really to satisfy my own curiosity.

Thanks,
Jim




©2024 climateprediction.net