Rally Feb. 20th-22nd

gd_barnes · 2009-02-16, 23:11

Let's do another rally; this time on port 4000 from Feb. 20th-22nd at the usual times.

David, can Free-DC make it 2 in a row vs. ROLP? Karsten may be able to help us this time. He said he now has home internet.

If interest is tepid, we could hold off until March. Another port could also be considered. I'm suggesting port 4000 because it has the smallest k's except for our 6k drive. Port 5000 would also be a possibility. If we did port 8000 again, we could nearly run out of work there. The sieving effort needs to catch up with that one before we do another rally on it.

Admin edit: We'll postpone a rally until later in March.

Gary

IronBits · 2009-02-16, 23:57

Unfortunately, we have a rally/dog pile/road kill of our own on Spinhenge@Home for a month that just started on the 14th.
Perhaps next time.

gd_barnes · 2009-02-17, 00:07

Quote:

Originally Posted by IronBits

Unfortunately, we have a rally/dog pile/road kill of our own on Spinhenge@Home for a month that just started on the 14th.
Perhaps next time.

lol; I didn't know you had regular "road kill" rallies. Funny. Now I don't feel so bad that we were your only road kill from the last one.

Perhaps we'll delay ours until after that then. There's nothing new and pressing at this moment. The last rally was good to really get the 8th drive moving.

Any other thoughts from anyone?

Gary

Mini-Geek · 2009-02-17, 00:14

Quote:

Originally Posted by gd_barnes

Any other thoughts from anyone?

I'm fine with any time but agree that we should wait until Free-DC can stomp us again.

Flatlander · 2009-02-17, 00:16

Quote:

Rally Feb. 20nd-22nd

lol
I did something similar in the sieving thread but quickly pretended it was deliberate.

AMDave · 2009-02-17, 00:29

I am not across the details and priorities of all of the drives, but recent posts lead me to think that we should have a mini chase to clean out IB-9000 and get it out of the way while the sieve for IB-8000 catches up.

Then we would have dried out 3 ports (thinking of kar_bon

)and be ready to have a big thumper on IB-8000 when everyone is available.

Which would then mean we can get on to some bigger, juicier numbers for the rest of the year.

Is that logical or did my brain just pass gas?

Brucifer · 2009-02-17, 01:01

yeah.... I'm of the same mind as AMDave. I'm working on 9000 anyway, and I'd rather just stay and see it knocked out too. :)

gd_barnes · 2009-02-17, 11:19

We'd actually like to keep them all going. For once (lol), I'm actually extremely pleased at the allocation of the resources on the project, mainly thanks to Bruce and Dave running a good number of cores on port 9000 at the moment. (Only a few people like to extensively work on non-top-5000 or double check work so it's great to have you guys around!)

Had you guys not moved some machines there, I would have had to move about 1-2 quads there to get it moving because it is a lot of work. But doing that would have meant I couldn't sieve as quickly as I am (3 quads on it right now) on k=1400-2000 for n=500K-1M.

There are 3 objectives at play here, of which all are being accomplished at the moment and so I hate to give up 1-2 of them to complete 1 of the others more quickly.

1. To find smaller primes to fill in the holes in the prime ranges.

2. To find top-5000 primes so as to remain competitive on top-5000.

3. To keep all k-ranges moving all (or most) of the time by having ongoing sieving on k=1003-2000/n=500K-1M. [Otherwise k=1400-2000 will stall for a long time at n=500K.]

By doing all of them at the same time, we accomplish all of them at once.

Dave, for higher n-ranges, we already have the 5th/6th/7th drives, which are at n=~640K, ~615K, and ~640K respectively going to n=1M. But I hear what you are saying on having a "big one" again on port 8000, which was great fun. But in order to do that, we need to complete sieving for the n=500K-600K portion of the n=500K-1M range in the sieving drive because by the time we have a rally, we're likely to be close enough to n=500K that we would run out of work.

As a general rule, we won't have rallies on non-top-5000 ranges unless plenty of people like that idea. I can easily be talked into a rally on port 9000 if enough people come out in favor of it. It's actually a little easier on the admins. Although there are more primes to be checked and listed, we don't have to worry about making sure that new people get the top-5000 primes submitted correctly, which is what takes the most admin time. We just end up making sure that they are all listed correctly in the drive thread and on Karsten's pages. That said, the primes and score boost that we get at top-5000 are a boon to the project when we run the rallies on top-5000 work so that is why I generally lean that way.

Gary

gd_barnes · 2009-02-17, 11:23

Quote:

Originally Posted by Flatlander

lol
I did something similar in the sieving thread but quickly pretended it was deliberate.

I have no idea what you are talking about.

Gary

Brucifer · 2009-02-17, 18:25

Well Gary, if things are rolling along okay now, then why change stuff with a rally? I'm content to run 9000 down, and it's going down okay. I sort of always thought that the purpose of a rally was to try and put special emphasis on something that needed it. Doesn't really sound like one is needed at present as things sit for the time being.

gd_barnes · 2009-02-18, 18:20

Quote:

Originally Posted by Brucifer

Well Gary, if things are rolling along okay now, then why change stuff with a rally? I'm content to run 9000 down, and it's going down okay. I sort of always thought that the purpose of a rally was to try and put special emphasis on something that needed it. Doesn't really sound like one is needed at present as things sit for the time being.

The purpose of a rally can be one thing or another or both. First, it can focus emphasis where needed but that is only the secondary consideration for having them. Second and the primary reason for having them is that it can bring huge resources and great exposure to the project for a couple of days.

For the last rally, we really weren't in bad shape on port 8000 but having the rally on that port with its large #'s of top-5000 primes brought monsterous resources to the project for 2 days and also helped bring some more permanent searchers to the project such as Brian (PCZ) and Dave (AMDave).

It is the project exposure and temporary large # of resources that makes the huge potential of rallies exciting. You could call them a large "advertisement" for the project, which is why David's servers and Dave's stats are so important to them. The competition aspect helps bring in the heavy-hitters.

Gary

2009-02-16, 23:11	#1
gd_barnes May 2007 Kansas; USA 7²×211 Posts	Rally Feb. 20th-22nd Let's do another rally; this time on port 4000 from Feb. 20th-22nd at the usual times. David, can Free-DC make it 2 in a row vs. ROLP? Karsten may be able to help us this time. He said he now has home internet. If interest is tepid, we could hold off until March. Another port could also be considered. I'm suggesting port 4000 because it has the smallest k's except for our 6k drive. Port 5000 would also be a possibility. If we did port 8000 again, we could nearly run out of work there. The sieving effort needs to catch up with that one before we do another rally on it. Admin edit: We'll postpone a rally until later in March. Gary Last fiddled with by gd_barnes on 2009-02-18 at 18:23

2009-02-16, 23:57	#2
IronBits I ♥ BOINC! Oct 2002 Glendale, AZ. (USA) 3·7·53 Posts	Unfortunately, we have a rally/dog pile/road kill of our own on Spinhenge@Home for a month that just started on the 14th. Perhaps next time. Last fiddled with by IronBits on 2009-02-16 at 23:57

2009-02-17, 11:19	#8
gd_barnes May 2007 Kansas; USA 7²×211 Posts	We'd actually like to keep them all going. For once (lol), I'm actually extremely pleased at the allocation of the resources on the project, mainly thanks to Bruce and Dave running a good number of cores on port 9000 at the moment. (Only a few people like to extensively work on non-top-5000 or double check work so it's great to have you guys around!) Had you guys not moved some machines there, I would have had to move about 1-2 quads there to get it moving because it is a lot of work. But doing that would have meant I couldn't sieve as quickly as I am (3 quads on it right now) on k=1400-2000 for n=500K-1M. There are 3 objectives at play here, of which all are being accomplished at the moment and so I hate to give up 1-2 of them to complete 1 of the others more quickly. 1. To find smaller primes to fill in the holes in the prime ranges. 2. To find top-5000 primes so as to remain competitive on top-5000. 3. To keep all k-ranges moving all (or most) of the time by having ongoing sieving on k=1003-2000/n=500K-1M. [Otherwise k=1400-2000 will stall for a long time at n=500K.] By doing all of them at the same time, we accomplish all of them at once. Dave, for higher n-ranges, we already have the 5th/6th/7th drives, which are at n=~640K, ~615K, and ~640K respectively going to n=1M. But I hear what you are saying on having a "big one" again on port 8000, which was great fun. But in order to do that, we need to complete sieving for the n=500K-600K portion of the n=500K-1M range in the sieving drive because by the time we have a rally, we're likely to be close enough to n=500K that we would run out of work. As a general rule, we won't have rallies on non-top-5000 ranges unless plenty of people like that idea. I can easily be talked into a rally on port 9000 if enough people come out in favor of it. It's actually a little easier on the admins. Although there are more primes to be checked and listed, we don't have to worry about making sure that new people get the top-5000 primes submitted correctly, which is what takes the most admin time. We just end up making sure that they are all listed correctly in the drive thread and on Karsten's pages. That said, the primes and score boost that we get at top-5000 are a boon to the project when we run the rallies on top-5000 work so that is why I generally lean that way. Gary Last fiddled with by gd_barnes on 2009-02-17 at 11:20

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
20th Test of primality and factorization of Lepore with Pythagorean triples	Alberico Lepore	Alberico Lepore	43	2018-01-17 15:55
Move the 20th (moving to endgame soon)	Dubslow	Game 1 - ♚♛♝♞♜♟ - Shaolin Pirates	10	2013-03-03 08:59
Server machine Aug. 20th crash and backups	kar_bon	No Prime Left Behind	29	2010-08-30 03:26
My own Rally	em99010pepe	No Prime Left Behind	4	2008-12-16 23:31
48 hr rally	em99010pepe	Sierpinski/Riesel Base 5	19	2006-11-08 00:52

2009-02-17, 00:29	#6
AMDave Jan 2006 deep in a while-loop 658₁₀ Posts	I am not across the details and priorities of all of the drives, but recent posts lead me to think that we should have a mini chase to clean out IB-9000 and get it out of the way while the sieve for IB-8000 catches up. Then we would have dried out 3 ports (thinking of kar_bon )and be ready to have a big thumper on IB-8000 when everyone is available. Which would then mean we can get on to some bigger, juicier numbers for the rest of the year. Is that logical or did my brain just pass gas?

2009-02-17, 01:01	#7
Brucifer Dec 2005 313 Posts	yeah.... I'm of the same mind as AMDave. I'm working on 9000 anyway, and I'd rather just stay and see it knocked out too. :)

2009-02-17, 18:25	#10
Brucifer Dec 2005 313 Posts	Well Gary, if things are rolling along okay now, then why change stuff with a rally? I'm content to run 9000 down, and it's going down okay. I sort of always thought that the purpose of a rally was to try and put special emphasis on something that needed it. Doesn't really sound like one is needed at present as things sit for the time being.