Top-5000 project scores table

gd_barnes · 2008-04-02, 20:35

This might be an interesting table to post once every 1-2 weeks to see the history of whether we are gaining or losing top-5000 'score' vs. other projects.

For the total score change by project since this thread was started see the total score change post.

The latest version of this table can be found at top 5000 project scores.

Code:

3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski
 137188 Seventeen or Bust
   4121 Riesel Sieve Project
   3639 Riesel Prime Search
   2451 PrimeGrid
   2236 Yves Gallot's GFN Search Project
   1583 Prime Internet Eisenstein Search
   1535 321search
   1245 The Prime Sierpinski Problem
   1206 GFN 2^17 Sieving project
   1019 12121 Search
    461 PrimeSearch
    324 Mat's Prime Search
    268 GFN 2^16 Sieving project
    248 Riesel Base 5
    204 15k*2^n-1 search
    162 Les GeneFermiers
     85 Twin Prime Search
     53 GFN 2^15 Sieving project
     48 GFN 2^14 Sieving project

Gary

Mini-Geek · 2008-04-08, 16:32

Code:

3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski
 137188 Seventeen or Bust
   4121 Riesel Sieve Project
   3705 Riesel Prime Search
   2448 PrimeGrid
   2236 Yves Gallot's GFN Search Project
   1583 Prime Internet Eisenstein Search
   1535 321search
   1245 The Prime Sierpinski Problem
   1206 GFN 2^17 Sieving project
   1019 12121 Search
    472 PrimeSearch
    324 Mat's Prime Search
    268 GFN 2^16 Sieving project
    248 Riesel Base 5
    204 15k*2^n-1 search
    162 Les GeneFermiers
     65 Twin Prime Search
     53 GFN 2^15 Sieving project
     48 GFN 2^14 Sieving project

RPS went up 66, PrimeGrid went down 3, PrimeSearch went up 11, and TPS went down 20. All else same as last time.

gd_barnes · 2008-04-08, 16:43

Quote:

Originally Posted by Mini-Geek

Code:

3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski
 137188 Seventeen or Bust
   4121 Riesel Sieve Project
   3705 Riesel Prime Search
   2448 PrimeGrid
   2236 Yves Gallot's GFN Search Project
   1583 Prime Internet Eisenstein Search
   1535 321search
   1245 The Prime Sierpinski Problem
   1206 GFN 2^17 Sieving project
   1019 12121 Search
    472 PrimeSearch
    324 Mat's Prime Search
    268 GFN 2^16 Sieving project
    248 Riesel Base 5
    204 15k*2^n-1 search
    162 Les GeneFermiers
     65 Twin Prime Search
     53 GFN 2^15 Sieving project
     48 GFN 2^14 Sieving project

RPS went up 66, PrimeGrid went down 3, PrimeSearch went up 11, and TPS went down 20. All else same as last time.

Thanks for the update Mini-Geek. RPS kicked butt this last week and we encountered somewhat of a dry spell.

Primegrid and the Prime Sierpinski Problem collectively found a monster prime that is still in the process of being confirmed. When confirmation is complete, the projects will split the score and it will score 1800+ for each project! :surprised That's the nature of the Sierp Prob project, though. They hadn't found a prime for nearly 18 months.

I think this next week should be better for us.

Gary

Mini-Geek · 2008-04-08, 16:55

Quote:

Originally Posted by gd_barnes

Thanks for the update Mini-Geek. RPS kicked butt this last week and we encountered somewhat of a dry spell.

Primegrid and the Prime Sierpinski Problem collectively found a monster prime that is still in the process of being confirmed. When confirmation is complete, the projects will split the score and it will score 1800+ for each project! :surprised That's the nature of the Sierp Prob project, though. They hadn't found a prime for nearly 18 months.

I think this next week should be better for us.

Gary

That's a lot of score, but it is a quite large prime. I think that all the non-Top 5K work, like sieving then LLRing of the double check drive, as well as simply the fact that we are testing higher ranges, have been slowing us. Plus, aren't some of our heavy hitters out right now?

gd_barnes · 2008-04-08, 17:11

Quote:

Originally Posted by Mini-Geek

That's a lot of score, but it is a quite large prime. I think that all the non-Top 5K work, like sieving then LLRing of the double check drive, as well as simply the fact that we are testing higher ranges, have been slowing us. Plus, aren't some of our heavy hitters out right now?

Yes, Beyond had some issues with a couple of his machines that he is looking into and is working on some other personal efforts right now. I just sent a pitch to him about our individual-k reservations for n>600K that I'm thinking he may be interested in.

Carlos is still working on quite a few manual ranges for us but has informed me that after he is done, he will stop prime searching in the near future in favor of another effort that he is very interested in. (darn)

Such is the ebb and flow of prime search projects. I looked at the results on the two servers and it does look like we're getting plenty of activity...much more than I expected so it's a combination of a lull in primes and a heavy hitter not working on the ranges right now. I know personally, I've searched in excess of 10000 pairs on port 5000 in the last 5-7 days with no prime. I've had 4-7 cores on it at all times plus 2-3 cores running manual ranges, also which have yielded no prime. (2 confirmed primes don't count) We're now past n=470K there, although there are still some manual ranges available for n=463K-470K. If those aren't taken within 7-10 days, I will pick them up after my current n=437K-440K and 460K-463K ranges are complete.

It also looks like I'll need to send Adam another range on port 300 within the next day or so. Even though I'm not working on that one at all, by my estimate, we have 2.5-3 days work in the queue and are past n=414K.

So...plenty of work is being done even though it's been a bit quiet here.

Gary

mdettweiler · 2008-04-08, 17:31

Quote:

Originally Posted by gd_barnes

<snip>

So...plenty of work is being done even though it's been a bit quiet here.

Yeah, that's just the nature of LLRnet and other automated systems. For example, take a look at Riesel Sieve and Seventeen or Bust. Both have lots and lots of work being done for them, yet their forums have hardly any action compared to NPLB, a comparatively small project. Why? Because almost all the action is automated.

So, as Gary said, even though some of the heavy hitters are gone and the action seems to have slowed a bit, we're still getting a lot of work done in LLRnet.

gd_barnes · 2008-04-16, 18:38

Code:

3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski
 137187 Seventeen or Bust
   4321 PrimeGrid
   4121 Riesel Sieve Project
   3738 Riesel Prime Search
   3128 The Prime Sierpinski Problem
   2236 Yves Gallot's GFN Search Project
   1583 Prime Internet Eisenstein Search
   1535 321search
   1206 GFN 2^17 Sieving project
   1019 12121 Search
    495 PrimeSearch
    324 Mat's Prime Search
    268 GFN 2^16 Sieving project
    253 Riesel Base 5
    206 15k*2^n-1 search
    162 Les GeneFermiers
     53 GFN 2^15 Sieving project
     48 GFN 2^14 Sieving project
     47 Twin Prime Search

Code:

Changes from last week:
+1883 The Prime Sierpinski Problem
+1873 PrimeGrid
+33   Riesel Prime Search
+23   PrimeSearch
+5    Riesel Base 5
+2    15k*2^n-1 search
-1    Seventeen or Bust
-18   Twin Prime Search

So...other than the huge prime found by Prime Sierpinski/PrimeGrid, we gained vs. all projects except RPS. Great work everyone! RPS is still the standard to beat...They have several large crunchers on their project.

I think we may be able to finally start closing the gap on RPS: I have parts for 6 quads on order that I hope to have built by the weekend that will replace the lesser firepower of 20+ machines that I can no longer access; hence the delay in my manual reservations. I may purchase 4 more quads in the next 2-3 months depending on how the money situation looks then.

We are still gaining slowly on RPS as far as # of primes but more of ours than theirs will start to drop off within the month so we'll have to crunch away to keep gaining.

One more note: Great work everyone in running the heck out of the port 8000 server! It looks like we may dry it before Friday; something I wasn't sure we would be able to do.

Gary

Mini-Geek · 2008-04-16, 19:23

I wish I could be helping PrimeSearch's score more, but I haven't found a prime in quite a while.

I'm hoping my current Drive 1 reservation of 900K n total (525K n left, just over 6000 candidates) will yield at least one prime. It's about a dual-core Athlon CPU month and expected about 1.8 primes, currently about 1.1 primes

Quote:

Originally Posted by gd_barnes

I think we may be able to finally start closing the gap on RPS: I have parts for 6 quads on order that I hope to have built by the weekend that will replace the lesser firepower of 20+ machines that I can no longer access; hence the delay in my manual reservations. I may purchase 4 more quads in the next 2-3 months depending on how the money situation looks then.

We are still gaining slowly on RPS as far as # of primes but more of ours than theirs will start to drop off within the month so we'll have to crunch away to keep gaining.

Wow, 6 quads. Nice. Must've been expensive.

Quote:

Originally Posted by gd_barnes

One more note: Great work everyone in running the heck out of the port 8000 server! It looks like we may dry it before Friday; something I wasn't sure we would be able to do.

Just thought I'd note that I've been running port 8000 since I saw the announcement, which was right after it went up (I think I was the first besides Anon to see it). I hope we can dry it out quickly. I'll hold off on gaming to help it speed up a bit.

gd_barnes · 2008-04-17, 10:23

Quote:

Originally Posted by Mini-Geek

I wish I could be helping PrimeSearch's score more, but I haven't found a prime in quite a while.

I'm hoping my current Drive 1 reservation of 900K n total (525K n left, just over 6000 candidates) will yield at least one prime. It's about a dual-core Athlon CPU month and expected about 1.8 primes, currently about 1.1 primes

Wow, 6 quads. Nice. Must've been expensive.

Just thought I'd note that I've been running port 8000 since I saw the announcement, which was right after it went up (I think I was the first besides Anon to see it). I hope we can dry it out quickly. I'll hold off on gaming to help it speed up a bit.

It's been 2+ weeks for me on a prime...I would guess 20K+ tests...so join the crowd. I've been hurt in the last week by losing access to 20+ machines (hence the quad purchase) but that still doesn't fully explain the dry spell I've been in. I have yet to find an n>400K prime on either drive. I switched over to drive 3 right before a spate of primes on drive 1. Oh well, it will all even out in the long run.

6 quads when done with everything...about $2800-3000 plus $300 to a friend to build and set them up with a single monitor, a monitor switching device, a remote desktop, and wireless adaptors in my cool basement. Of course I have to learn Linux to avoid spending $600 on Windows for all of them. But I got a book on it and it looks like it shouldn't be hard to learn.

Newegg has great prices.

Gary

Flatlander · 2008-04-17, 12:10

Quote:

Originally Posted by gd_barnes

6 quads when done with everything...about $2800-3000 plus $300 to a friend to build and set them up with a single monitor, a monitor switching device, a remote desktop, and wireless adaptors in my cool basement. Of course I have to learn Linux to avoid spending $600 on Windows for all of them. But I got a book on it and it looks like it shouldn't be hard to learn.
Gary

Commandment 11:
Thou shalt not covet thy fellows prime crunchers.

Somebody hide my credit card!

Beyond · 2008-04-17, 13:13

From the Prime Pages;

The PrimeGrid/TPS project has once again uncovered a weakness in the Prime Page's submission system with their large number of submissions that have the same number of digits. The code for the submission page looks at the number of digits and is not accurate enough currently to spot primes that have a rank too small to make the list when submitted, but have the same number of digit as the 5000th prime.

This will be caught later as the list is reranked (they are the primes on the list whose entrance rank is greater than 5000), and I will manually remove them, but there are several difficulties.

* Comparing general numbers this size can be time consuming, especially when the server is loaded and many primes are submitted at once. (Comparing TPS numbers is easy, but the solution must work for all forms.)
* I must make sure

looks like some changes on the way that may effect the scoring.

2008-04-02, 20:35	#1
gd_barnes May 2007 Kansas; USA 3³·5·7·11 Posts	Top-5000 project scores table This might be an interesting table to post once every 1-2 weeks to see the history of whether we are gaining or losing top-5000 'score' vs. other projects. For the total score change by project since this thread was started see the total score change post. The latest version of this table can be found at top 5000 project scores. Code: 3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski 137188 Seventeen or Bust 4121 Riesel Sieve Project 3639 Riesel Prime Search 2451 PrimeGrid 2236 Yves Gallot's GFN Search Project 1583 Prime Internet Eisenstein Search 1535 321search 1245 The Prime Sierpinski Problem 1206 GFN 2^17 Sieving project 1019 12121 Search 461 PrimeSearch 324 Mat's Prime Search 268 GFN 2^16 Sieving project 248 Riesel Base 5 204 15k2^n-1 search 162 Les GeneFermiers 85 Twin Prime Search 53 GFN 2^15 Sieving project 48 GFN 2^14 Sieving project Gary Last fiddled with by gd_barnes on 2008-07-24 at 20:36 Reason: add link for total score change*

2008-04-16, 18:38	#7
gd_barnes May 2007 Kansas; USA 10100010011011₂ Posts	Code: 3892015 Great Internet Mersenne Prime Search by Woltman & Kurowski 137187 Seventeen or Bust 4321 PrimeGrid 4121 Riesel Sieve Project 3738 Riesel Prime Search 3128 The Prime Sierpinski Problem 2236 Yves Gallot's GFN Search Project 1583 Prime Internet Eisenstein Search 1535 321search 1206 GFN 2^17 Sieving project 1019 12121 Search 495 PrimeSearch 324 Mat's Prime Search 268 GFN 2^16 Sieving project 253 Riesel Base 5 206 15k2^n-1 search 162 Les GeneFermiers 53 GFN 2^15 Sieving project 48 GFN 2^14 Sieving project 47 Twin Prime Search Code: Changes from last week: +1883 The Prime Sierpinski Problem +1873 PrimeGrid +33 Riesel Prime Search +23 PrimeSearch +5 Riesel Base 5 +2 15k2^n-1 search -1 Seventeen or Bust -18 Twin Prime Search So...other than the huge prime found by Prime Sierpinski/PrimeGrid, we gained vs. all projects except RPS. Great work everyone! RPS is still the standard to beat...They have several large crunchers on their project. I think we may be able to finally start closing the gap on RPS: I have parts for 6 quads on order that I hope to have built by the weekend that will replace the lesser firepower of 20+ machines that I can no longer access; hence the delay in my manual reservations. I may purchase 4 more quads in the next 2-3 months depending on how the money situation looks then. We are still gaining slowly on RPS as far as # of primes but more of ours than theirs will start to drop off within the month so we'll have to crunch away to keep gaining. One more note: Great work everyone in running the heck out of the port 8000 server! It looks like we may dry it before Friday; something I wasn't sure we would be able to do. Gary Last fiddled with by gd_barnes on 2008-04-16 at 18:50

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Best msieve poly scores	VBCurtis	Msieve	181	2021-05-01 16:24
4000 < k < 5000	otutusaus	Riesel Prime Data Collecting (k*2^n-1)	5	2012-03-07 20:01
5000 < k < 6000	justinsane	Riesel Prime Data Collecting (k*2^n-1)	26	2010-12-31 12:27
As Math Scores Lag, a New Push for the Basics -- New York Times	ewmayer	Soap Box	14	2006-11-17 09:40
Top-5000 List	edorajh	Riesel Prime Search	17	2006-03-28 21:57

2008-04-17, 13:13	#11
Beyond Dec 2002 2²·3·37 Posts	From the Prime Pages; The PrimeGrid/TPS project has once again uncovered a weakness in the Prime Page's submission system with their large number of submissions that have the same number of digits. The code for the submission page looks at the number of digits and is not accurate enough currently to spot primes that have a rank too small to make the list when submitted, but have the same number of digit as the 5000th prime. This will be caught later as the list is reranked (they are the primes on the list whose entrance rank is greater than 5000), and I will manually remove them, but there are several difficulties. * Comparing general numbers this size can be time consuming, especially when the server is loaded and many primes are submitted at once. (Comparing TPS numbers is easy, but the solution must work for all forms.) * I must make sure looks like some changes on the way that may effect the scoring.