Hy guys!
I\'ve got a serious problem: I can\'t find a way to put in the \"Running\" state my Coupled Model :-(
I\'ve joined the project three-four days ago with the TruXoft Boinc client on my P4 3.0GHz with HT. On this pc are running also other projects.
Since my pc is always full of work from the other project, the manager do the scheduling using \"earlies-deadline-first\" policy. And even if the resource share has been set @1500 for climate and @100 for the other project, there is no chance that climate got the focus!
The only solution is to suspend all but Climate, but I can\'t follow my Boinc manager 24/7, and even in this way I lose the HT, \'cause I\'ve got only one Climate model!!
Have you ever experienced something similar?
Any sugggestion?

Thanks a lot!
- Sorry for my english, I\'m italian!! -

The new models require a BOINC version of 5.*
Is the TruXoft version from this series, or from the 4.* series?

It\'s from 5.2.13 version!

---

In that case, just let BOINC handle things.
It\'ll get around to running climate in due course.
If you try to micro manage things, you\'ll just make it worse, because BOINC has to keep \'re-arranging it\'s time table\' every time you fiddle.
The climate model will take well over a thousand hours, so there\'s no rush.

That\'s ok, I\'ll try this strategy!
I\'ve crunched til 0.530% in about 23 hours (3.72 s/TS).
I hope to reach soon 0.625% to gain my first Climate Trickle!!

I will give you updates on this topic!

Thank a lot, Les!

---

You need to lower your connect to number, so the cache will go down and then Boinc will not need to go into earlest-deadline-first mode.

---

BOINC Wiki

I agree, Jim K!
I\'m emptying my cache and I\'ve lowered the preference you mention.
And the TruXoft client doesn\'t help, cause it ask more work w.r.t. the standard client.
I hope to solve the puzzle before sunday!

Thanks for the hint!

---

The cache went down...

07/04/2006 15.05.49||Rescheduling CPU: application exited
07/04/2006 15.05.49|rosetta@home|Computation for result TRUNCATE_TERMINI_2chf__416_778_0 finished
07/04/2006 15.05.49||Allowing work fetch again.
07/04/2006 15.05.49||resuming round-robin CPU scheduling.
07/04/2006 15.05.49|climateprediction.net|Restarting result hadcm3lb_4yh8_05019288_1 using hadcm3lb version 508

and Climate resumed the run!!
Everything is clear... I\'ve got to take low the cache by fine-tuning the \"Connect to network about every\" preference.

And now... Engine at max speed toward my first trickle!!

Thanks again Jim and Les!
See you soon!

---

And this is my first trickle!!
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/trickle.php?resultid=5067306

Just one thing:
The progression has been easy \'til 0.53%;
then the first application crash... wu rolled back to 0.52% and restarted;
arrived at 0.60%, 0.62%, 0.63% three other crash, always rolled back to 0.60%.
At last... the first trickle...
Since then, everything works fine and now i\'m at 0.75%.

Obviously if I wasn\'t at my desk the client was still there, still blocked!
Please note that on my WinXpSP2 P4 3.0GHz HT no other program was in execution when every crash occurred, nor the Climate graphical interface!

So why 4 crash before the trickle?
Is it normal?

---