Riesel base 16 - team drive #2

[tnerual]

5307*16^76522-1 is prime! Time : 559.0 sec.

[mdettweiler]

71K-74K complete, no primes. lresults attached.

[tnerual]

74-80 done only one prime

[tnerual]

15458*16^81157-1 is prime! Time : 333.0 sec.

[kar_bon]

range 56k-71k complete (one prime already posted)

taking 85k-87.5k

[mdettweiler]

Taking 87.5K-90K for the LLRNet server.

[gd_barnes]

Originally Posted by Anonymous

Taking 87.5K-90K for the LLRNet server.

Pardon my techno-stupidity here. I'm assuming this means that you'll be running it through your machine but getting feeds from the server?

That is, I will show you as reserving it and you would get credit for any top-5000 prime.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Pardon my techno-stupidity here. I'm assuming this means that you'll be running it through your machine but getting feeds from the server?

That is, I will show you as reserving it and you would get credit for any top-5000 prime.


Gary

No, all that means is that the file is going to be fed into the LLRnet server, which will hand out the individual candidates just like a regular team drive (except that it's all automated). Any top-5000 primes would be credited to whoever's computer reported the result to the server. (Carlos will send me the results files, which are pretty much regular lresults files, but with added time/date and user information. I can strip that out to make it look like a regular lresults file--should be easy with a quickie perl script.)

Thus, you should show the file as reserved by "LLRnet". (The same goes for any k's we may dedicate to the LLRnet server in the future.)

[gd_barnes]

Originally Posted by Anonymous

No, all that means is that the file is going to be fed into the LLRnet server, which will hand out the individual candidates just like a regular team drive (except that it's all automated). Any top-5000 primes would be credited to whoever's computer reported the result to the server. (Carlos will send me the results files, which are pretty much regular lresults files, but with added time/date and user information. I can strip that out to make it look like a regular lresults file--should be easy with a quickie perl script.)

Thus, you should show the file as reserved by "LLRnet". (The same goes for any k's we may dedicate to the LLRnet server in the future.)

Holy sh-moly! I'd say I'll get crackin' on sieving Riesel base 16. Jasong offered some sieving but I haven't seen him around in a while. I'll get it started and see if I can coordinate with him on helping.

Anon, instead of feeding the LLRnet server any more Riesel files, we've got a whole slew of Sierp base 16 files and nobody has reserved any work on that side since I restarted the drive except for me. I have them available all the way up to n=200K, although only posted to n=120K.

OK, how do these people know what project to report the primes under? CRUS still gets 100% credit; correct?


Gary

[mdettweiler]

I've split Gary's original post into multiple parts for easier replying:

Originally Posted by gd_barnes

Holy sh-moly! I'd say I'll get crackin' on sieving Riesel base 16. Jasong offered some sieving but I haven't seen him around in a while. I'll get it started and see if I can coordinate with him on helping.

Okay, sounds good!

Anon, instead of feeding the LLRnet server any more Riesel files, we've got a whole slew of Sierp base 16 files and nobody has reserved any work on that side since I restarted the drive except for me. I have them available all the way up to n=200K, although only posted to n=120K.

Well, we'd have to first let the LLRnet server run dry if we were going to do that...an LLRnet server can't have more than one type of numbers (i.e. numbers with a different header in the NewPGen format file) queued up, just like regular LLR...but it still would work OK, I would think, especially since we have few enough Riesel candidates in the server that it shouldn't take long to run them dry. But, you're right, right now it would be better to do Sierp. Base 16 in the LLRnet server.

OK, how do these people know what project to report the primes under? CRUS still gets 100% credit; correct?

Simple--in the thread that I'm going to post about the LLRnet server, it will include complete instructions for running the LLRnet client--including how to report primes. Primes are reported just as they would be in a regular team drive--remember, LLRnet is simply an automated team drive system. They are still reported as being found by LLR, too--there isn't a separate prover code for LLRnet, since it's just a frontend to LLR. Thus, LLRnet makes no difference in how the primes are reported.

[gd_barnes]

I have put 2 cores sieving all 27 remaining k's for Riesel base 16 from n=100K-200K (17 in the drive). I had sieved up to 25K-200K initially before breaking off n=25K-100K so I'm starting from where I left off the first piece of P=600G.

My estimate for optimal sieve is 1.1T. I can get to 1T in 2-1/2 days and am right now running P=600G-800G and P=800G-1T on the 2 cores. I'll see if Jasong is available to run P=1T-1.1T. If not, tomorrow, I'll have a 3rd core available for that. Either way, I should be ready to post files by Friday.

Optimal may be closer to 1T if we knock out 2-3 k's before hitting n=100K here.

If we're comfortable that the LLRNet server is stable and running well, if you guys want to feed a chunk of the rest of the files for Riesel base 16 to the LLRNet server before switching to Sierp, go for it.


Gary