mersenneforum.org LLRnet/PRPnet rally June 4th-6th
 Register FAQ Search Today's Posts Mark Forums Read

 2010-06-06, 18:21 #23 Sloth     Mar 2006 1368 Posts ending at the defined time vs the extra 3 hours will not make a difference on where I end up :P
2010-06-06, 19:00   #24
gd_barnes

May 2007
Kansas; USA

100111111101002 Posts

Quote:
 Originally Posted by Sloth ending at the defined time vs the extra 3 hours will not make a difference on where I end up :P

Well, what I was thinking is where the TEAMS end up.

I haven't even looked at any team or individual counts. I'm just trying to do the right thing for LLRnet vs. PRPnet folks since it was only the PRPnet server that went down for a while.

Lennart and Sloth, can I get a vote on it one way or another? Also, if Max and henryzz want to vote, that would be good too. I think Max has already voted.

2010-06-06, 19:25   #25
henryzz
Just call me Henry

"David"
Sep 2007
Cambridge (GMT/BST)

166516 Posts

Quote:
 Originally Posted by gd_barnes Well, what I was thinking is where the TEAMS end up. I haven't even looked at any team or individual counts. I'm just trying to do the right thing for LLRnet vs. PRPnet folks since it was only the PRPnet server that went down for a while. Lennart and Sloth, can I get a vote on it one way or another? Also, if Max and henryzz want to vote, that would be good too. I think Max has already voted.
I think it probably should be extended although i have removed my machine as it wouldn't have been on at the point the problem was.

2010-06-06, 19:59   #26
Lennart

"Lennart"
Jun 2007

25×5×7 Posts

Quote:
 Originally Posted by gd_barnes Well, what I was thinking is where the TEAMS end up. I haven't even looked at any team or individual counts. I'm just trying to do the right thing for LLRnet vs. PRPnet folks since it was only the PRPnet server that went down for a while. Lennart and Sloth, can I get a vote on it one way or another? Also, if Max and henryzz want to vote, that would be good too. I think Max has already voted.
I does not matter for me :) I will soon start moving my cores to other work.

But it was fun

Did we not find any prime on the rally ?

Lennart................................................................

2010-06-06, 20:34   #27
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

792 Posts

Quote:
 Originally Posted by Lennart I does not matter for me :) I will soon start moving my cores to other work. But it was fun Did we not find any prime on the rally ? Lennart................................................................
You found one prime, but you're right...I don't think we got any more. Seems a shame to have such a dearth of primes for a rally, though I suppose it's just as good for the project as a whole when you consider that every composite we found brings us that much closer to reaching the actual primes, wherever they are.

2010-06-06, 20:41   #28
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

792 Posts

Quote:
 Originally Posted by gd_barnes Well, what I was thinking is where the TEAMS end up. I haven't even looked at any team or individual counts. I'm just trying to do the right thing for LLRnet vs. PRPnet folks since it was only the PRPnet server that went down for a while. Lennart and Sloth, can I get a vote on it one way or another? Also, if Max and henryzz want to vote, that would be good too. I think Max has already voted.
Assuming we stick with the original end time, the final totals are as follows:
Code:
Team                       Total
Raiders of the Lost Primes 14543
PrimeSearchTeam            14168
Team-Goobee.org             2823
AMD Users                    856
Very close, indeed. Given Lennart's sustained output in the hour after the end of the rally (20:00 on the port report) and assuming that he cranks out at least 2/3 or so of that amount for 21:00, extending the rally by 2 hours would make just enough difference for PrimeSearchTeam to win by a hair.

Actually, though, if you think about it, it's probably better for it to go this way (with ROLP winning by a very slim margin) in terms of the project as a whole: this way Lennart can go back to his team and report that they lost by a hair, and that they should bring tons of resources next time to help him out and make up the difference!

In my earlier post I mentioned that the precedent we have set in the past for situations like this is to extend the rally for the amount of time lost due to downtime; however, to clarify that further, what we always did was float it around and play it case-by-case based on what the participants felt. Thus if everyone is fine with it staying like this, it would still not be particularly unfair given how we've done this in the past.

Last fiddled with by mdettweiler on 2010-06-06 at 20:43

 2010-06-06, 21:05 #29 Flatlander I quite division it     "Chris" Feb 2005 England 31×67 Posts If it's been extended by 3 hours then ... Calling all of ROLP. Everyone jump on the PRPNet server!!!
 2010-06-06, 21:40 #30 gd_barnes     May 2007 Kansas; USA 22·2,557 Posts No, like I said earlier. We're leaving it on the honor system for people who were on the LLRnet server during the outage to not do the PRPnet server. Let's extend it. This is really cool. I think Lennart planned it that way That is just enough to outdo ROLP; although we don't know just yet. That's what I had been hoping for with one of the rallies. A battle right down to the end. It's too bad we had the outage. Then we could have done hourly updates to see the ebb-and-flow, just like a Lakers / Celtics NBA finals game! I gotta talk Ian into putting at least a couple of quads on one of these or maybe I'll have to pick up a couple of them myself. :-)
2010-06-06, 21:52   #31
Lennart

"Lennart"
Jun 2007

46016 Posts

Quote:
 Originally Posted by gd_barnes No, like I said earlier. We're leaving it on the honor system for people who were on the LLRnet server during the outage to not do the PRPnet server. Let's extend it. This is really cool. I think Lennart planned it that way That is just enough to outdo ROLP; although we don't know just yet. That's what I had been hoping for with one of the rallies. A battle right down to the end. It's too bad we had the outage. Then we could have done hourly updates to see the ebb-and-flow, just like a Lakers / Celtics NBA finals game! I gotta talk Ian into putting at least a couple of quads on one of these or maybe I'll have to pick up a couple of them myself. :-)

This was what I would like to see a match, race or call it what you like.

I am not sure PST will take ROLP if you extend the race. But it is close

Lennart

2010-06-07, 03:31   #32
gd_barnes

May 2007
Kansas; USA

100111111101002 Posts

Quote:
 Originally Posted by mdettweiler Assuming we stick with the original end time, the final totals are as follows: Code: Team Total Raiders of the Lost Primes 14543 PrimeSearchTeam 14168 Team-Goobee.org 2823 AMD Users 856 Very close, indeed. Given Lennart's sustained output in the hour after the end of the rally (20:00 on the port report) and assuming that he cranks out at least 2/3 or so of that amount for 21:00, extending the rally by 2 hours would make just enough difference for PrimeSearchTeam to win by a hair. Actually, though, if you think about it, it's probably better for it to go this way (with ROLP winning by a very slim margin) in terms of the project as a whole: this way Lennart can go back to his team and report that they lost by a hair, and that they should bring tons of resources next time to help him out and make up the difference! In my earlier post I mentioned that the precedent we have set in the past for situations like this is to extend the rally for the amount of time lost due to downtime; however, to clarify that further, what we always did was float it around and play it case-by-case based on what the participants felt. Thus if everyone is fine with it staying like this, it would still not be particularly unfair given how we've done this in the past.

Even though we extended the PRPnet part by 3 hours, like you said, 2 hours was enough because Lennart started pulling off midway through the 2nd hour. Here is what appears to be the final stats:

Code:
PST        14168 + 384 + 244 + 24 =    14820
ROLP       14543 + 18 + 30 + 24 + 10 = 14625
Team-Goobee 2823 + 81 + 75 + 68 =       3047
AMD Users    856
WOW!! What a battle and congrats to PST for a hard fought victory!

Max, can you work with Dave on the rally stats page to extend port 9000 only to 22:00 on June 6th? If done correctly, I think the above should be exactly correct unless I missed something somewhere.

Last fiddled with by gd_barnes on 2010-06-07 at 03:33

2010-06-07, 03:39   #33
gd_barnes

May 2007
Kansas; USA

22×2,557 Posts

Quote:
 Originally Posted by Lennart This was what I would like to see a match, race or call it what you like. I am not sure PST will take ROLP if you extend the race. But it is close Lennart
Looks like you did! Thanks everyone for participatiing in the rally! While we were short on primes, we got a lot of pairs processed. It appears that we did 32K-33K pairs vs. 37K pairs in the last one but the last one had a fair portion of a smaller fftlen so not a big difference. I hope this will turn out to be the start of a good rivalry between ROLP and PST for the next few rallies!

I'd like to do another in mid-late July but before deciding, I'd like to guage some interest at some of the other forums. With the mid-summer heat in the northern hemisphere, some people shut part or all of their machines down.

Max,

Please go ahead and load an n=2K range into port 9000. Doing it tonight would be best. We'll hold off a while yet on port 3000. I'll be pulling my machines off over the next few hours so it will be quite a while before we finish out that range.

In doing the above, we'll have a day or 2 to decide if we want to load
another n=3K range into port 9000 to make it the usual n=5K range; based on the ongoing load on the server.

Gary

 Similar Threads Thread Thread Starter Forum Replies Last Post mdettweiler No Prime Left Behind 55 2011-04-25 09:35 mdettweiler No Prime Left Behind 48 2011-01-12 10:14 mdettweiler No Prime Left Behind 33 2010-12-24 19:16 mdettweiler No Prime Left Behind 88 2010-09-09 12:50 mdettweiler No Prime Left Behind 67 2008-06-23 15:32

All times are UTC. The time now is 01:39.

Thu Oct 29 01:39:58 UTC 2020 up 48 days, 22:50, 1 user, load averages: 1.33, 1.44, 1.57