f/f george wrote:

On Fri, 25 Jun 2004 08:10:19 +0000 (UTC), "RD" <rd@dragon.net> wrote:

Hi Ed,
"Ed" <@> wrote in message
news:1088149899.7681.0@spandrell.news.uk.clara.net...

"RD" <rd@dragon.net> wrote:

only crunching 2 WU per day then the 20 WU per day cruncher is
going to be waiting a damm long time for his/her credits:-/

I thought so too.  Looking at the specs of the other 2 machines
crunching my WUs is very scary.  I am running a P II 233 Win2000
box whilst all
of the others are P4s in the 2.4 to 3.2 GHz range.  But it has been
interesting to see that I'm not actually letting the side down as I
thought I might.  My first 3 WUs go something like this ("S" =
sent, "W" = waiting; my machine is "C"):

A    B    C
S    W    S
W   W    S
S    S    W (82% done)

Pretty surprising, huh?  I haven't got credit for the first 2 WUs
because my P II 233 is still waiting for the P4s.  Perhaps they
have already done their WUs, but it is still sitting on their
disks.  Or they have used a large cache...

Anythings possible:-) But as others have posted it seems to level
out over time. However I thought the cache size was determined by
boinc after the speed test.. Can one manually set the cache size?

Cheers,
          RD

Yes go to the Boinc page and then to your account, scroll down to
preferences, general, and click on view/edit, you can change the
minimum and maximum days to cache work units on your machine. A number
of .5 is okay to use if you so choose.
I have mine set at 1 day minimum and 2 days maximum, but I have a
cable connection and am always "on" the net.
Currently Berkeley has a maximum of 50 units per computer as the max
size of the cache. That may change as the program settles down.
You don't want to big of a cache because of the "report deadline" date
that is found for each workunit under the "work" tab. If you return
any unit after that date you WILL NOT get credit for it, AND Boinc
will automatically send it out to others computers for crunching.
The more that happens, the longer you and I have to wait to get credit
for our on-time finished unit.

I hope that sometime in the future Berkeley allows different cache sizes for
different machines. I have a machine that is always on, a machine that is on
12-14 hours a day and a couple of machines that are turned on for, on
average, 24 hours a week. I'd hate to have them crunching units that are
worthless, both for the science and the credits. At the moment I can see
that at least one of my machines is going to be spinning it's wheels,
running at 100% CPU when I fire it up, just to return expired units.
--
~misfit~