My computer has a Celeron 2.0G CPU, and 512MB RAM.

There is a page <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/cpu.html">here</a> that gives some comparisons for different machines. You should look for several with similar hardware to yours, as sometimes the information for a particular machine may be misleading, or there may be special factors influencing the speed. The final column shows how many days are needed to complete the task running all the time.

it looks about a month (last column is days to do 1 workunit):

Intel(R) Celeron(R) CPU 2.00GHz Pentium 1 Microsoft Windows 2003 1698.0 3761.0 1535.52 976.56 3.345 30.11

Thanks for all.

One month is long time. Especially there is another project work unit waiting for crunching...

&gt; One month is long time. Especially there is another project work unit waiting
&gt; for crunching...

BOINC is designed to run in the background of your machine for months or years. The projects use redundant cycles on the CPU, and you get credit for the same amount of work irrespective of which project you are doing.

Comparing CPDN with SETI or P@H in terms of time taken to do a single WU is to miss the point. SETI is concerned with analysing vast amounts of data split into very small chunks, sent out as individual WUs. Berkeley could just as easily split it up into bigger chunks which would take you a month, but it is not what people are used to and would take some time to download. CPDN distributes a full model run, modelling the earth's climate over a total of 45 years; it is necessarily big but is the most efficient way of doing it. Each to his own; personally I like the idea of doing a complete experiment rather than little bits shared with others, but others do not.

If you prefer not to tie your machine up with just one project for a long time, BOINC will now let you share between projects (now just SETI and CPDN, but P@H will soon join) and switch between them.

&gt; If you prefer not to tie your machine up with just one project for a long
&gt; time, BOINC will now let you share between projects (now just SETI and CPDN,
&gt; but P@H will soon join) and switch between them.

This is actually quite nice, because as often as S@H goes down, you always have something you can work on during the outages. :)

A single CPDN WU can put you in the gravy for weeks or even months depending on your CPU speed. Having one of these bad boys in your cache virtually ensures your PC will never sit idle regardless of how often the other projects go down.

<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>

So the progress bar is wildly uneven, and the initial estimates are insanely high (a couple of orders of magnitude)?

I've now got more than 45 hours invested into my first workunit (congrats, you guys are benefitting from the ongoing confusion at seti and the predictor@home people), and it *still* shows over 1100 hours of work remaining.

This is, frankly, absurd; find a way to cut the workunits down into smaller pieces!

(Windows 2k, AMD 2800+ processor, 512MB ram)

5.1 secs/timestep does seem rather slow for that CPU - I assume it is a Barton XP 2800+ running at 12.5x166.7 for 2083MHz..? My current slowest CP-Boinc machine, at 2.86 secs/timestep, is 'Oony' - her Barton is currently running at 12.5x176 for 2200MHz - that is only roughly 6% faster than your 2800+, again assuming your CPU <i>is</i> a Barton, I would expect your 2800+ to process at just over 3 secs/timestep, not 5.1...

BTW: As other folks have already stated, there is <i>no way</i> to split the climate model 'work_units' into small pieces without excessive bandwidth usage. Check 'I/O Read Bytes' &amp; 'I/O Write Bytes' in your task manager to see what I mean - eg. 'Alison' has currently been running CP-boinc for 12h17m, she has 'I/O Read Bytes'=26GB &amp; 'I/O Write Bytes'=6GB..!

<a href="http://www.nmvs.dsl.pipex.com/"><img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=6&amp;team=off&amp;trans=off"></a>

&gt; This is, frankly, absurd; find a way to cut the workunits down into smaller
&gt; pieces!

It is, frankly, not absurd. When I first joined this project (i.e. my main qualification was "I ran SETI@Home for years! :-) I felt the same way. Although now I have come to enjoy watching the whole planet "evolve" through the simulation. Think how disjointed it would be if one day you were doing a winter's day in 1843, then the next day you dowloaded another workunit to do a summer day in 2063? Other than the logistical problems, i.e. do you really want to download/upload 10MB of restart files every day etc?

In the "grand scheme of things" and thanks to BOINC pre-emption, if you run BOINC for awhile, does it really matter if you did 20 "workunit widgets" from Project A and 2 from Project B as long as the credits all make sense, you see what you have done etc?