Page 1 of 1

WU point values

Posted: Mon Jun 01, 2009 6:19 am
by alpha754293
Where would I be able to look up the point values for various WUs?

Re: WU point values

Posted: Mon Jun 01, 2009 6:45 am
by rpmouton
alpha754293 wrote:Where would I be able to look up the point values for various WUs?
here is the link for currently running WUs:

http://fah-web.stanford.edu/psummary.html

regards,

Re: WU point values

Posted: Mon Jun 01, 2009 6:47 am
by alpha754293
rpmouton wrote:
alpha754293 wrote:Where would I be able to look up the point values for various WUs?
here is the link for currently running WUs:

http://fah-web.stanford.edu/psummary.html

regards,
So, am I to interpret this that all WUs within a given project get the same point value regardless of their RCG? (Just (double-)checking.)

Re: WU point values

Posted: Mon Jun 01, 2009 6:54 am
by rpmouton
Alpha754293,

That is my understanding and observation seems to validate it as well.

PS I can think of several candidates for the forgotten city in MI. Which is yours?

Re: WU point values

Posted: Mon Jun 01, 2009 7:05 am
by 7im
Yes, all WUs from the same project get the same total points.

However, in rare cases, some projects are known to have variable length work units, and so the total points may vary.

Also, not all projects will earn a consistent PPD. The WUs from some projects will get faster as the generations progress. Some slow as the generations progress. Pande Group typically sets the original benchmark to account for this... i.e. setting the original PPD higher so the final work units will still score at or close to the benchmark.

Re: WU point values

Posted: Mon Jun 01, 2009 7:07 am
by alpha754293
7im wrote:Yes, all WUs from the same project get the same total points.

However, in rare cases, some projects are known to have variable length work units, and so the total points may vary.

Also, not all projects will earn a consistent PPD. The WUs from some projects will get faster as the generations progress. Some slow as the generations progress. Pande Group typically sets the original benchmark to account for this... i.e. setting the original PPD higher so the final work units will still score at or close to the benchmark.
Yea, I'm crunching the numbers now for the two "-smp 4" clients right now. There's a slight variation between WUs of the same project, but it isn't by much. So far, it averages out to around 10 minutes on a 14.7 hour run.

Re: WU point values

Posted: Mon Jun 01, 2009 5:56 pm
by bruce
alpha754293 wrote:There's a slight variation between WUs of the same project, but it isn't by much. So far, it averages out to around 10 minutes on a 14.7 hour run.
That's about 1%. The general rule-of-thumb quoted around here is that you can expect variations up to 5% before anybody really starts worrying about anything. (Plus there's almost always variations in what else your computer might have been doing during that 14.7 hours.)

Re: WU point values

Posted: Mon Jun 01, 2009 6:51 pm
by alpha754293
bruce wrote:
alpha754293 wrote:There's a slight variation between WUs of the same project, but it isn't by much. So far, it averages out to around 10 minutes on a 14.7 hour run.
That's about 1%. The general rule-of-thumb quoted around here is that you can expect variations up to 5% before anybody really starts worrying about anything. (Plus there's almost always variations in what else your computer might have been doing during that 14.7 hours.)
Yea, I wasn't sure how much variation to expect, but I've noticed that there's some, but I don't think that it is anything to worry about. Even an oh-so-dramatic 10-20% variation is acceptable by my standards because the configuration (of the run) can change, or there's something inherently more complicated that it is working on.

For the time being, there isn't much else that's running in the background other than normal system processes. I didn't bother trying to shut down most of the rest of the crap cuz I just don't have time to go through them and to determine the bare minimum essentials.

Samba would be about the only thing on the system that gets a bit of mileage out of because FahMon reads the client data over that; but most everything else is done via a telnet session in (which is pretty lightweight) and I only really have to look at it if there's something wrong.

Re: WU point values

Posted: Mon Jun 01, 2009 6:59 pm
by 7im
As long as were on the topic, there are also things called folding events that can affect the PPD of a particular workunit, and not all work units.

See also: http://fahwiki.net/index.php/Folding/Aggregation_Event

Re: WU point values

Posted: Mon Jun 01, 2009 7:03 pm
by alpha754293
7im wrote:As long as were on the topic, there are also things called folding events that can affect the PPD of a particular workunit, and not all work units.

See also: http://fahwiki.net/index.php/Folding/Aggregation_Event
(internal joke) ha ha...I almost seem like I actually have a clue as to what I'm talking about.

Didn't realize that when I said that "it might get more complicated" that it's actually true. NICE. lol.