I switched one computer over to 4.X yesterday and joined CPDN. I had been running SETI and Predictor on that machine and did not detatch them, assuming that they would be joining in the BOINC 4 fun soon.

This afternoon I noticed that the client was requesting more work, even though it has 730 hrs. left to do on the WU I started yesterday. If SETI had got thing going Friday as planned, I would be adding a SETI WU which would not begin crunching until well after its deadline had passed.

Why is BOINC requesting more work? Is there anything I can do short of detatching everything else when I start processing a CPDN model?

I am running on a Pentium 4, 2.6 GHz, XP maching. My options page is set to 1 to 2 days of work.

Carlos

4.xx uses pre-emptive scheduling, so other projects may be worked on while doing a long WU like CPDN. BOINC will swap projects every hour. I believe it is asking for more work so it can balance your resource debt.

<a href="http://www.boinc.dk/index.php?page=user_statistics&amp;project=cpdn&amp;userid=355"><img border="0" height="80" src="http://355.cpdn.sig.boinc.dk?188"></a>

&gt; 4.xx uses pre-emptive scheduling, so other projects may be worked on while
&gt; doing a long WU like CPDN. BOINC will swap projects every hour. I believe it
&gt; is asking for more work so it can balance your resource debt.
&gt;
&gt; <a> href="http://www.boinc.dk/index.php?page=user_statistics&amp;project=cpdn&amp;userid=355"&gt;<img> border="0" height="80" src="http://355.cpdn.sig.boinc.dk?188"&gt;</a>
&gt;

Thanks. That answers the question.

&gt; BOINC will swap projects every hour.

Ah .. that answers a question I was just about to post .. as I see that Seti has been hogging the CPU for the last 35 minutes or so and CPDN appears to be suspended. A 1 hour swap cycle makes sense.

Would there be much benefit in changing the preferences to keep the applications in memory at all times? I am presuming that if the app has to be reloaded (and restarted) there would be some overhead required to restart and resynch its calculations.

if you can afford the 50MB of memory "suspended" it's actually better to do that than to have it at the default setting of quitting on a preempt. You won't lose any work between checkpoints, and there won't be as much startup overhead etc.

PS -- I just got my first 4.x SETI workunit so I guess they got things up at Berkeley today!

&gt; if you can afford the 50MB of memory "suspended" it's actually better to do
&gt; that than to have it at the default setting of quitting on a preempt.

Ok .. thanks Carl ..

I have changed my preferences in both projects to "suspend" and will give that a whirl.

&gt; I have changed my preferences in both projects to "suspend" and will give that
&gt; a whirl.

It's not actually necessary to change this in both projects. The general preferences are BOINC global, and will migrate from project to project. Any setting changed in general prefs will affect projects. You change it on one account page, and after the other project talks to the servers, you will notice it changes for that project as well. :)

<a href="http://www.boinc.dk/index.php?page=user_statistics&amp;project=cpdn&amp;userid=355"><img border="0" height="80" src="http://355.cpdn.sig.boinc.dk?188"></a>