Riesel base 6 - team drive #4 - EIGHT OR BUST! - Page 35

mdettweiler · 2014-08-26, 23:39

I step away from the forum for a week and a new R6 prime is found just in time to drop off the new-posts list when I come back?! Figures...

At the risk of sounding redundant (which is surely excused by the significance of this find), huge congrats!

Gary, do you have the odds of prime numbers on hand for R6? Off the top of my head, I seem to recall a somewhat <1 expected primes in the 1M-2M range, which would continue this base's track record of performing extremely well. This would seem to bode well for the remaining k=1597...

As I mentioned in the PM I just sent, yes, the credit for this is correct. The vast majority of CRUS's sieves are conducted in-house with the srsieve family of programs. Only the power-of-2 bases could potentially be different (some of them can make use of the PrimeGrid sieve files, others not, depending on the specific ranges and bases...best ask Gary on a case-by-case basis for those, I've long since lost track).

Regarding the N+1 test, don't quote me on this, but I think that's an artifact of the way LLR implements the N+1 method. In PFGW, the N+1 is a special "primality proof mode", and is much slower than a standard PRP; LLR manages to do them both at once, but, from the looks of it, roughly the same process may be going on behind the scenes, with the initial test only confirming probable primality (perhaps to a higher degree than standard PRP?), and the second test being needed to confirm absolute proof*.

In any event, the top-5000 site runs the full test, so we know for sure now (doubly) that it's prime: the first test done by PRPnet would have been the full thing (LLR with N+1 and the Lucas step); and the top-5000 site ran a full N+1 with PFGW.

*Well, within the bounds of reliability of the physical computer systems we're working on...you know. Just being pedantic.

mdettweiler · 2014-09-01, 20:08

Wow, at the current rate of processing we're going to reach n=2M in just a couple of days! Thanks Lennart for the extended boost - it's good to see this base getting its wind again after that last prime.

Thinking back to our previous discussions on future plans for port 1400, I believe the ultimate plan is to load it with the even/odd base-2 k's when the sieving drive for those finishes. The big question, of course, is: how much longer is the sieving drive going to take?

Right now it's completely stalled.

I'm willing to shift some resources over to it when the current R6 work finishes up. It's been a long time since I did any sieving, but I should warn everyone, by myself it'll take me anywhere from 5-10 months to finish with the resources I have available. I've just got one slow-ish AMD quad and a fast-ish Intel dualcore (the latter only running part-time) to commit; of the other two boxes I have, one is running a 32-bit OS, and the other is difficult to access (and thus stays on NPLB port 2000 full time to provide a steady trickle of top-5000 primes

).

Clearly, unless we want to leave CRUS without a PRPnet testing drive for the better part of a year, either a) I'm going to need some help with the sieving, or b) we need to find another interim effort for port 1400.

Unless anyone's got other preferences, my suggestion would be to grab one or two low bases from the one-k thread that have sieve files available and push those upward while we're waiting on the base 2 sieve. Personally, I am quite interested in seeing our oldest efforts (bases <=32) brought to a proof, and there are a few ripe candidates (in a similar vein to R6). Looking at the list now, I see S9 at n=1.5M with a sieve file to n=2M, and S27 at n=800K with a sieve file to 1M. There's also S22 and S26, each with 2 k's at n=500K.

If others are not particularly interested, I might grab one of these as a personal effort and redirect my CRUS resources there, but I would really prefer to work as part of a team effort. It's much more fun.

Lennart · 2014-09-01, 20:30

Quote:

Originally Posted by mdettweiler

Thinking back to our previous discussions on future plans for port 1400, I believe the ultimate plan is to load it with the even/odd base-2 k's when the sieving drive for those finishes. The big question, of course, is: how much longer is the sieving drive going to take?

Right now it's completely stalled.

You can load base 2 !

You know how to remove candidates by using factorfiles with prpadmin ?

So that's no problem if we start sieving.

Lennart

mdettweiler · 2014-09-02, 00:57

Quote:

Originally Posted by Lennart

You can load base 2 !

You know how to remove candidates by using factorfiles with prpadmin ?

So that's no problem if we start sieving.

Lennart

Yes, I'm familiar with how to remove factor files with prpadmin; however, I was under the impression that our sieve is still quite a ways from optimal depth even for the first part of the range. Gary, please correct me if I'm wrong on this, and I can break off a section to load right away.

gd_barnes · 2014-09-02, 07:59

Quote:

Originally Posted by mdettweiler

Yes, I'm familiar with how to remove factor files with prpadmin; however, I was under the impression that our sieve is still quite a ways from optimal depth even for the first part of the range. Gary, please correct me if I'm wrong on this, and I can break off a section to load right away.

Let's get it sieved up to P=650T for breaking off n=2M-5M. I can't switch my machines over for another week. If anyone can help with the sieving drive, that will be great.

gd_barnes · 2014-09-03, 07:27

I see that the drive is now complete to n=2M minus one straggling pair. Thank you everyone for your contributions!

What we could do is load the remaining part of S9 up to n=2M into port 1400. The only issue is that we don't really have an official miscellaneous "PRPnet servers for CRUS". We would kind of need to start one if we wished to do various misc. efforts with PRPnet servers since S9 is not really a team drive.

It has been suggested and I agree that we should have two PRPnet servers running at a time for the project. After I get back in a week and after we complete the base 2 even-odd-k-n sieving effort, maybe that is something that we could look into. Perhaps we could have both that effort and the S9 effort going in two different servers.

mdettweiler · 2014-09-03, 08:43

Hmm, good point. I suppose we could simply declare it a "PRPnet mini-drive", but perhaps that's overkill for a single k.

(Still, in terms of Nash weight S9 has actually got a tad more heft to it than R6 k=1597...)

If there's interest in a second PRPnet server for "miscellaneous efforts", I could restart the long-defunct port 1300 server (with a clean database and the latest prpserver version).

MyDogBuster · 2014-09-03, 09:10

I think a second server doing stuff like S9 to 2M is a good idea. There must be more candidates than that similar to S9. I'll scour the pages for more. Good idea to load more than 1 just in case a prime is found.

Edit: Aha - S22 with 2 k's remaining from 500K to 1M. Sieve file at 50T.
Worst case scenario is we could throw any 1ker out there from the recommended list..

mdettweiler · 2014-09-03, 15:00

Quote:

Originally Posted by MyDogBuster

I think a second server doing stuff like S9 to 2M is a good idea. There must be more candidates than that similar to S9. I'll scour the pages for more. Good idea to load more than 1 just in case a prime is found.

Edit: Aha - S22 with 2 k's remaining from 500K to 1M. Sieve file at 50T.
Worst case scenario is we could throw any 1ker out there from the recommended list..

Indeed. Maybe we could start it off with S9 and S22? Of course, then the question is, which one to do first: S22 will have rather smaller candidates, so given prpserver's various sorting options we couldn't do them together; however, we could do them in either order.

I suppose I could always fiddle with the source code to make it randomly pick one of the two bases whenever it gives a candidate to a client...just looked at the code now and I think I see how it could be done.

MyDogBuster · 2014-09-03, 21:45

Quote:

Indeed. Maybe we could start it off with S9 and S22? Of course, then the question is, which one to do first: S22 will have rather smaller candidates, so given prpserver's various sorting options we couldn't do them together; however, we could do them in either order.

I suppose I could always fiddle with the source code to make it randomly pick one of the two bases whenever it gives a candidate to a client...just looked at the code now and I think I see how it could be done.

If you use the a,l sort option in the config for the server, then whichever base you load first will get done first. It will do them in loaded order, smallest length first.

gd_barnes · 2014-09-04, 07:48

Ian, Max, or Mathew, can you copy the "PRPnet servers for NPLB" first post from the NPLB project over to here and obviously call it "PRPnet servers for CRUS"? Then tweak it for this project and sticky it. Port 1300 will be for S9 (misc.) efforts and port 1400 will be for the future base 2/4 even-odd-k-n effort.

Port 1300 sounds good. I would suggest doing S9 to n=2M first. At P=50T, I don't think S22 is sieved far enough yet for n=500K-1M. I can make that determination when I get back.

2014-08-26, 23:39	#375
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	I step away from the forum for a week and a new R6 prime is found just in time to drop off the new-posts list when I come back?! Figures... At the risk of sounding redundant (which is surely excused by the significance of this find), huge congrats! Gary, do you have the odds of prime numbers on hand for R6? Off the top of my head, I seem to recall a somewhat <1 expected primes in the 1M-2M range, which would continue this base's track record of performing extremely well. This would seem to bode well for the remaining k=1597... As I mentioned in the PM I just sent, yes, the credit for this is correct. The vast majority of CRUS's sieves are conducted in-house with the srsieve family of programs. Only the power-of-2 bases could potentially be different (some of them can make use of the PrimeGrid sieve files, others not, depending on the specific ranges and bases...best ask Gary on a case-by-case basis for those, I've long since lost track). Regarding the N+1 test, don't quote me on this, but I think that's an artifact of the way LLR implements the N+1 method. In PFGW, the N+1 is a special "primality proof mode", and is much slower than a standard PRP; LLR manages to do them both at once, but, from the looks of it, roughly the same process may be going on behind the scenes, with the initial test only confirming probable primality (perhaps to a higher degree than standard PRP?), and the second test being needed to confirm absolute proof. In any event, the top-5000 site runs the full test, so we know for sure now (doubly) that it's prime: the first test done by PRPnet would have been the full thing (LLR with N+1 and the Lucas step); and the top-5000 site ran a full N+1 with PFGW. Well, within the bounds of reliability of the physical computer systems we're working on...you know. Just being pedantic. Last fiddled with by mdettweiler on 2014-08-26 at 23:42 Reason: typo in smiley code

2014-09-01, 20:08	#376
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	Wow, at the current rate of processing we're going to reach n=2M in just a couple of days! Thanks Lennart for the extended boost - it's good to see this base getting its wind again after that last prime. Thinking back to our previous discussions on future plans for port 1400, I believe the ultimate plan is to load it with the even/odd base-2 k's when the sieving drive for those finishes. The big question, of course, is: how much longer is the sieving drive going to take? Right now it's completely stalled. I'm willing to shift some resources over to it when the current R6 work finishes up. It's been a long time since I did any sieving, but I should warn everyone, by myself it'll take me anywhere from 5-10 months to finish with the resources I have available. I've just got one slow-ish AMD quad and a fast-ish Intel dualcore (the latter only running part-time) to commit; of the other two boxes I have, one is running a 32-bit OS, and the other is difficult to access (and thus stays on NPLB port 2000 full time to provide a steady trickle of top-5000 primes ). Clearly, unless we want to leave CRUS without a PRPnet testing drive for the better part of a year, either a) I'm going to need some help with the sieving, or b) we need to find another interim effort for port 1400. Unless anyone's got other preferences, my suggestion would be to grab one or two low bases from the one-k thread that have sieve files available and push those upward while we're waiting on the base 2 sieve. Personally, I am quite interested in seeing our oldest efforts (bases <=32) brought to a proof, and there are a few ripe candidates (in a similar vein to R6). Looking at the list now, I see S9 at n=1.5M with a sieve file to n=2M, and S27 at n=800K with a sieve file to 1M. There's also S22 and S26, each with 2 k's at n=500K. If others are not particularly interested, I might grab one of these as a personal effort and redirect my CRUS resources there, but I would really prefer to work as part of a team effort. It's much more fun. Last fiddled with by mdettweiler on 2014-09-01 at 20:09

2014-09-03, 09:10	#382
MyDogBuster May 2008 Wilmington, DE 101100100100₂ Posts	I think a second server doing stuff like S9 to 2M is a good idea. There must be more candidates than that similar to S9. I'll scour the pages for more. Good idea to load more than 1 just in case a prime is found. Edit: Aha - S22 with 2 k's remaining from 500K to 1M. Sieve file at 50T. Worst case scenario is we could throw any 1ker out there from the recommended list.. Last fiddled with by MyDogBuster on 2014-09-03 at 09:25

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Riesel base 16 - team drive #2	gd_barnes	Conjectures 'R Us	213	2014-02-26 09:35
Sierp base 63 - team drive #5	rogue	Conjectures 'R Us	146	2011-04-20 05:12
Sieving drive Riesel base 6 n=1M-2M	gd_barnes	Conjectures 'R Us	40	2011-01-22 08:10
Sieving drive Riesel base 6 n=150K-1M	gd_barnes	Conjectures 'R Us	27	2009-10-08 21:49
Riesel base 3 - mini-drive I	gd_barnes	Conjectures 'R Us	199	2009-09-30 18:44

2014-09-03, 07:27	#380
gd_barnes May 2007 Kansas; USA 101·103 Posts	I see that the drive is now complete to n=2M minus one straggling pair. Thank you everyone for your contributions! What we could do is load the remaining part of S9 up to n=2M into port 1400. The only issue is that we don't really have an official miscellaneous "PRPnet servers for CRUS". We would kind of need to start one if we wished to do various misc. efforts with PRPnet servers since S9 is not really a team drive. It has been suggested and I agree that we should have two PRPnet servers running at a time for the project. After I get back in a week and after we complete the base 2 even-odd-k-n sieving effort, maybe that is something that we could look into. Perhaps we could have both that effort and the S9 effort going in two different servers.

2014-09-03, 08:43	#381
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 1869₁₆ Posts	Hmm, good point. I suppose we could simply declare it a "PRPnet mini-drive", but perhaps that's overkill for a single k. (Still, in terms of Nash weight S9 has actually got a tad more heft to it than R6 k=1597...) If there's interest in a second PRPnet server for "miscellaneous efforts", I could restart the long-defunct port 1300 server (with a clean database and the latest prpserver version).

2014-09-04, 07:48	#385
gd_barnes May 2007 Kansas; USA 101×103 Posts	Ian, Max, or Mathew, can you copy the "PRPnet servers for NPLB" first post from the NPLB project over to here and obviously call it "PRPnet servers for CRUS"? Then tweak it for this project and sticky it. Port 1300 will be for S9 (misc.) efforts and port 1400 will be for the future base 2/4 even-odd-k-n effort. Port 1300 sounds good. I would suggest doing S9 to n=2M first. At P=50T, I don't think S22 is sieved far enough yet for n=500K-1M. I can make that determination when I get back.