When will NPLB be the top project...

[MooooMoo]

...in terms of number of top 5000 primes found?

I made a poll on this at the beginning of February when the difference between NPLB and the top project was about 150 primes. The most popular option was that NPLB would be #1 sometime in summer, although a few people did say that this project wouldn't reach the top until the end of the year or later.

Now, the difference between the top two projects are only 32 primes (923 vs. 891). So, let's get this prediction contest going - anyone have some guesses as to when NPLB be at the top? Also, how many top 5000 primes will we have found at that time?

My guess is May 28th. NPLB will have 942 primes, and the runner-up will have 941.

edit: Gary, are you planning on doing a rally soon?

[gd_barnes]

I would like to have a rally in either 2 weeks or 4 weeks. If anyone has any opinions on the timing of the next one, go ahead and state them if you want to.

[IronBits]

Ahem, you have a Free-DC NPLB BBQ going on for the entire month of May.
Do you really feel you need to have a another rally running while that's going on?

[gd_barnes]

LMAO; OK maybe not.

ROLP, it looks like we need to get busy!

Max, I'll copy those results files off to my main machine so that we can get the manually done results loaded into the servers. That should help.

[gd_barnes]

We will blow right by RPS in 15-20 days. Estimate: May 20th.

Next goal: To be the top non-BOINC project in terms of prime score, which would mean 4th place behind GIMPS, SOB, and PrimeGrid. We've been gaining extremely fast on the 3 projects directly above us for months now, 2 of which are (were) BOINC projects (RieselSieve is dead). Estimate: July 31st, 2010.

We're closing in and I can smell an NPLB BBQ!


Gary

[Flatlander]

It's funny. I used to quite like BBQs, but now ...

My estimate: May 14th.

[gd_barnes]

Originally Posted by MyDogBuster

Lets see. As far as number of primes, Jun 7th, 2009. As far as score, I'm a little more optimistic, Feb 1st, 2010.

I think a month long BBQ is enough punishment for a while. Us "old timers", as David likes to refer to us, need more recovery time the older we get.

Lets see, if I go out and spend my grandsons' and my 2 sisters inheritances, just how many I7's would I have to buy to keep from getting my butt flamed once a month?

With Bruce changing to ROLP, if you can put your full gammit of machines on a rally, we could give Free-DC a pretty good run for their money...after May that is. lol Especially if I decide to pick up a new I7 myself. Editor note: Gary hasn't decided on that yet.


Gary

[IronBits]

If we can get the auto-registration and team affiliation thing worked out and get it listed on the DC Vault, more folks would come.
After this May run by Free-DC, I wouldn't worry much about their output in the long run as most folks will go back to doing their favorite 'Boinc' thing.

Speaking of Boinc, we need to get the sieving part of this project on Boinc.

Where are the stats for all the Sieving??????????
Where are the stats for double checking????????

[Brucifer]

Sieving goes to boinc and you can count me out of any more sieving. As for the BBQ's, I'm sorta in the same frame of mind on them that you are Flatlander. As for rally's, definitely not in for one this month, think I'll just lay low for the month and finish the sieving I'm working on.

[gd_barnes]

Originally Posted by Flatlander

It's funny. I used to quite like BBQs, but now ...

My estimate: May 14th.

Originally Posted by Brucifer

Sieving goes to boinc and you can count me out of any more sieving. As for the BBQ's, I'm sorta in the same frame of mind on them that you are Flatlander. As for rally's, definitely not in for one this month, think I'll just lay low for the month and finish the sieving I'm working on.

You guys both missed my meaning of BBQ here. I wasn't referring to an NPLB rally with Free-DC BBQing everyone. I was referring to a BBQ where NPLB BBQ's RPS for 1st place on # of primes!! That's what we're doing right now. We're only 16 behind as of this posting and will now likely beat my May 20th estimate. It's 933-917 at this moment so we've outgunned them 26-10 since Mike's 1st post here. That's in just 4-1/2 days with all of the primes being n>590K (and a large majority n>600K)! If we do the same in the next 4-1/2 days, it will be 943-943 early on May 12th. Therefore, Mike's estimated count of 942-941 when we pass them is looking like an excellent one.

We'll wait a while on another rally; at least until late June...but to me, the main point of a rally is to bring exposure to the project; a big advertisement if you will. I still would like to encourage people to participate and if you don't like the competition aspect, just ignore the rally scores/stats and Free-DC or whomever bragging about winning. I think it's funny. If Free-DC BBQ's ROLP and everyone else, let 'em have the bragging rights. It's a boon to the project! Actually, that had been my hope all along and why we kept egging them on after beating then 2-3 straight times. They then smoked us the last 2, which is what I would expect from such a large group. I had been disappointed before that. The fact is we've gotten some great well-healed long-term searchers here as a result of the rallies. I belive that PCZ and Vaughan both came here and stayed as a result of a rally. That's what it's all about. Personally, I really enjoy rallies for just that reason.

As for BOINC on sieving. Bruce, the only reason that I would ever consider BOINC for any aspect of any project that I am on is to do work that no one here wants to do. Let me give two examples:
1. At CRUS, doing testing on just about any base for n>~1M. Tests that are not base 2 take 5+ times as long for the same size test. So instead of taking ~1000 secs., it would likely take > 5000 secs. (nearly 1.5 hours) to do one test. But you wouldn't get any more score credit on the top-5000 site because that is based on the size of prime, not what base is being tested.
2. At NPLB, sieving some monster range that would take so long, few would want to dedicate their machines to it. Example: Sieving k<10000 for all n=1 to 5 million. I'm not sure regular computer RAM or srsieve could handle such a monster range but I'm sure both could handle something very huge that few would be interested in dedicating long-term resources to.

Regardless, the point should be clear. BOINC would only be used for extremely annoying huge tasks that are otherwise relatively thankless and that few people would consider doing. Double-checking might be another example.

That said, if the BOINC group would want some sharing of top-5000 primes because of helping us sieve, I wouldn't go for it at all. Sieving is only 5-7% of any effort. Splitting primes is not worth it unless they actually do the prime tests like in the CRUS example.


Gary

[vaughan]

Originally Posted by gd_barnes

I believe that PCZ and Vaughan both came here and stayed as a result of a rally. That's what it's all about. Personally, I really enjoy rallies for just that reason.
Gary

That's partially correct.