Subject: Re: BOINC Progress meter calibration
From: "Stratcat" <none@no.org>
Date: 28/06/2004, 20:12
Newsgroups: alt.sci.seti

"~misfit~" <misfit61nz@yahoo-mung.co.nz> wrote in message
news:IZIDc.4727$LT3.185146@news.xtra.co.nz...
Stratcat wrote:
"Jan Schejbal" <use_reply-to@ungueltig.invalid> wrote in message
news:2k8iroF18n0mtU1@uni-berlin.de...
Hi!
hey will all have been completed well before the deadline

I am not running my PC 24/7, so they won't necessarily.

Jan


Then you may consider changing your cache size to reflect an accurate
estimate of the work you'll crunch in your preferred time frame.

Ahhh, but Strat my friend, what do you do when you are in my situation? I
have one machine that runs 24/7, one that runs 14/7 and one that just gets
turned on for a few hours whenever it's needed. All on the same account.
All
have been given work for 1-3 days.

Berkeley needs to allow us to set different prefences for each machine.
I'm
waiting to see if my machine that is only on occaisonally actually
crunches
the units it has once they're expired (if they do) or wether it'll
communicate with the server and dump them. I don't like tha fact that
someone, posibly even myself, will be waiting for the confirmational
results
for those units to get claimed credit and it won't happen from that
machine.

I'd like to bump up my WU cache to make allowances for server outages. I
know that at this early stage we're all chomping at the bit waiting for
credit confirmation but later a few days here or there won't mean nearly
as
much. However if I set my preferences to, say, seven days, there will be a
hell of a lot of WUs being given to the machines I have that aren't on so
often that have to be re-sent to another cruncher at a later date.

You could say "Forget about crunching with the machine that is only on for
around 24 hours a week" but you know I don't want to do that. If I wasn't
on
such a limited income I'd happily leave them all on 24/7. However the
Invalid's benefit doesn't pay enough for that.
--
~misfit~


Absolutely right, Misfit. I'm using machines that are similar in
performance, and all running & connected 24/7 (DSL), so it didn't occur
to me.

Excellent point, and even though it was brought up in my team forum, I
completely forgot. We definitely need a "machine flexible" caching method.
Otherwise, I suppose the only other choice, is to set cache to your lowest
OP machine - NOT a very good option.

Or, run seperate accounts - again not a great option, either.

BTW - Do you have a cable provider that offers both ISP services, and
basic anolog CTV???
-- Strat