um?

[Supp]

One notebook of mine, P4 by chance, is collecting dust in closet, I'll fire it up

Anybody else ?

[EDIT] Hm...what about adding it to Front Page Rankings ?

[rsbriggs]

We've had quite a turn out so far - 16 boxen!

Intel Pentium II I       :    3
AMD Athlon               :    1
Intel Celeron            :    1
Unspecified type         :    4
---------------------- -------
TOTAL, uniquely named :      16

How do you set the CPU type? My P4 shows as being "unspecified type".

Team info is here

[Beyond]

Originally posted by rsbriggs
We've had quite a turn out so far - 16 boxen!

Intel Pentium II I       :    3
AMD Athlon               :    1
Intel Celeron            :    1
Unspecified type         :    4
---------------------- -------
TOTAL, uniquely named :      16

How do you set the CPU type? My P4 shows as being "unspecified type".

Team info is here

You don't, the client does it automagically (there is a problem with reading the processor info on some P4's, nothing to worry about, all unspecified are P4's).

[magnav0x]

Not sure if I got it set up right, but I added my P4. BTW where is the statistics page for FreeDC? I can't seem to find it at www.mersenne.org

[willy1]

Magnav0x -

You're showing in the stats. rsbriggs posted a stats link a couple of posts back.

[magnav0x]

I don't know if there is a processor advantage between the assignment types. Basically you have 3 choices:

How exactly do we choose between the 3 choices? I didn't see any way to do this in the client.

[rshepard]

TEST MENU
---------

The PrimeNet menu choice lets you configure the type of work you get
from the server. The "Use PrimeNet..." option can be turned on to
switch from the manual method to the automatic method.
The "Request whatever..." box should be left checked. However, if you
are running a slow computer and don't mind waiting several months for
a single Lucas-Lehmer test to complete OR you are running a faster
computer and would rather do factoring, then uncheck this box and
choose a different type of work to do. The "Always have at least
this many days of work queued up" value should be changed based on
how often you connect to the Internet. As long as you connect at least
once in the given time period, mprime will have an uninterrupted stream
of work. However, the program will not checkout more than 20 exponents
no matter what this value is set at.

[magnav0x]

Is there a scoring difference between these units. I mean, are you going to get more movement in the rank from running a double-check rather than factoring a prime? Rank is determined by CPU time right?

[magnav0x]

I know some people are against crunching GIMPS for ourself, because they think it's useless, but here is my reasoning in crunching GIMPS for Free-DC.

Our mathmatics 2 in DC Repository stinks! You may think it's pointless to be last place in GIMPS. In all actuality it hurts us more to not participate in the project at all.

Taken from the DC Repository FAQ:

The number below the official rank is the local percentage of work (LPW) completed. Black text indicates no change from the previous day, green text indicates an increase and red text indicates a decrease. The LPW is calculated by taking the ratio between the team's "WU's" for the project and the sum of all DC Repo teams' "WU's" for the project and scaling by 100. A project "WU" is the measure used to rank teams. Note that this may or may not be a "work unit".

As you can see, even being last place in rankings would help us more than not having a ranking at all. My goal with GIMPS is not to move free-dc to the top of the project. We all know that would take a very long time and large ammount of resources.

[rsbriggs]

The only question is, do they consider "Free-DC" (used for most projects) and "FreeDC" (used for the math projects, like SOB and GIMPS) to be the same team?

[magnav0x]

I already e-mailed the guy at DC Repository stating our team name would be different in GIMPS

[rshepard]

Is there a scoring difference between these units. I mean, are you going to get more movement in the rank from running a double-check rather than factoring a prime? Rank is determined by CPU time right?

LL tests get more credit, because of the difference in CPU time required. You'll get results"on the board" quicker with factoring tests, but they aren't worth as much individually as an LL test (the exact ratio escapes me at the moment)

[magnav0x]

Which one is LL? I see:

Request 10,000,000 digit numbers to test
Request Mersenne numbers to run priliminary test
Request Mersenne numbers to double check
Request Mersenne numbers to factor

[rshepard]

I'm not at my Windows box, but are you sure that second item isn't "primality test"? if so , that's the LL test. The 10,000,000 digit is also LL, but is a looooooooooong run.

[magnav0x]

That's all I have to choose from. This is the beta client that was recently released.

[rshepard]

The second choice is the LL test-- I remoted to one of my boxes at work and confirmed that is correct.

[magnav0x]

Ok, thanks a lot

[Beyond]

Originally posted by magnav0x
Is there a scoring difference between these units. I mean, are you going to get more movement in the rank from running a double-check rather than factoring a prime? Rank is determined by CPU time right?

Workunit selection is done here.



10,000,000 digit to test are hugh compared to the standard LL Test, but could concievable win you some serious money.

Primality testing gives you the best chance of finding a prime and is the default workunit assigned to machines above 500MHz

DoubleChecks are just that, verifing runs on 1st time primaly test.

Factoring test are used to eliminate any exponent's from the much longer primatlity testing ( sieving more or less )

[ECL]

So now that you're off and rolling, and FreeDC is ranked 30024, don't you think it's time to suck up to Beth133? He's ranked 36. A combination would place FreeDC at 36th in the project (with a fair shot at 35 or 34), and second in Gimps in the DC Repository. That would raise FreeDC's Math2 score a bit, methinks.

There may be other FreeDC-friendly types wandering around between 36 and 30024.

[magnav0x]

So now that you're off and rolling, and FreeDC is ranked 30024, don't you think it's time to suck up to Beth133? He's ranked 36. A combination would place FreeDC at 36th in the project (with a fair shot at 35 or 34), and second in Gimps in the DC Repository. That would raise FreeDC's Math2 score a bit, methinks.

Wow, wouldn't that be a big turn of events

I wouldn't even know how to go about contacting him though

[rsbriggs]

Where are you coming up with the 300024 place number? I couldn't find us on the stats list which cut off in the 29,000's...

[magnav0x]

Yes I wish I was able to find out where exactly we can find our ranking at any given time.

[rshepard]

You can use this:
http://mersenne.org/cgi-bin/primenet...?UserID=FreeDC

[rsbriggs]

Ahh - that works. I suspect that we'll be moving up somewhat in the list shortly...

[rshepard]

looks like 4-6 days for the first results to start hitting-- after that they are nicely distributed, with results coming in every few days.

[magnav0x]

I found another link that shows ALL teams. If anyone is interested I could whip up a script that would keep track of our progress, overtake dates, etc. I don't know how fun they would be to watch, with such slow return rates though. I don't think I would include individual machines in the stats, just the team a whole. Up to you guys.


http://mersenne.org/ips/topproducers.shtml

[rsbriggs]

Looks to me like Team_Prime_Rib and TeamPrimeRib appear as separate teams, one at # 1, the other at 3,000 something.

We're in a thousand-way tie for 30025th place.....

It would be nice to track the upward progress for a while - we should be able to move up thousands of places quite rapidly, at least at the beginning.

[rsbriggs]

Hokey-smokes Batman! We've doubled again since yesterday.
(Where'd NineBall GET all those boxen?)


Lucas-Lehmer testing : 26
Factoring only : 20
Double-checking LL : 13
---------------------- -------
TOTAL : 59

[PCZ]

Did it matter what tests you do ?

Do all the points show up in the same database or is there 3 different stats.
It's a bit confusing.

[rshepard]

The "overall production" is scored like so:

Results are credited first by Lucas-Lehmer CPU time, then by factoring CPU time, next the number of exponents for which a factor was found, and finally by the number of exponents Lucas-Lehmer tested

They also maintain a database ranked just by factoring but it is of less interest than the overall production.

[willy1]

Originally posted by rsbriggs
Hokey-smokes Batman! We've doubled again since yesterday.
(Where'd NineBall GET all those boxen?)


Lucas-Lehmer testing : 26
Factoring only : 20
Double-checking LL : 13
---------------------- -------
TOTAL : 59

Not really -

That's the number of exponents assigned, not number of boxen. It looks like some people downloaded a couple of months's work for each box.


------- Machines Assigned to PrimeNet -------

Intel Pentium III : 17
AMD Athlon : 2
Intel Celeron : 2
Unspecified type : 8
---------------------- -------
TOTAL, uniquely named : 29



Of course, I presume someone could name all their boxes the same....


In any case, good work !!!! It will be nice to see the team rise through the standings as the work starts to be returned.

[MickFoley]

Welcome to the show, glad to see you are getting some people interested allready.


heres to you guys joing in on the fun with us.

MickFoley
Team Prime Rib

[Darkness Productions]

Just as an FYI, I've tossed my work machine (p3 500) in the fray. Not much, but like they say over at Ars, every box counts.

[Beyond]

Originally posted by Darkness Productions
Just as an FYI, I've tossed my work machine (p3 500) in the fray. Not much, but like they say over at Ars, every box counts.

Thanks DP, any and all machines, users, crunchers welcomed!

[ECL]

I would be derelict in my traditional role of wet blanket if I didn't mention this, but those of you who started crunching Gimps in order to improve our standing in the DC Repository and are factoring are wasting your time. Gimps has two measurements, LL P90 years and Factoring P90 years. Meremortal is only counting LL P90 years, which is why Free-DC is not yet even on the board over there.

On the other hand, if you're crunching Gimps factors for the sheer industrial hell of it, then carry on as you were. Nothing to see here. Go about your business.

Perhaps Meremortal could be prevailed upon to either add a Gimps Factoring category, or add the two numbers together for a combined score.

[FoBoT]

so what choice on the gimps setup do we select to contribute to

LL P90 years

?

how do i tell my box to not do "factoring " (the thing that won't contribute or whatever)

[rshepard]

Select Test==>Primenet and uncheck the "request whatever work makes sense" box, then check anything EXCEPT "Request Mersenne numbers to factor"

[magnav0x]

21544. FreeDC 0.678 1 0.996 2 2443.94


Finaly some headway

First LL test recieved apparently

[rshepard]

And there was much rejoicing

[FoBoT]

Originally posted by rshepard
Select Test==>Primenet and uncheck the "request whatever work makes sense" box, then check anything EXCEPT "Request Mersenne numbers to factor"

so if i picked the double checking thingy, i did good?

i thought i read somewhere that the double check option was relatively quick, so i picked that