n=1.7M and n=3.322M Twin-Sophie record attempts - Page 4

gd_barnes · 2023-01-29, 22:02

I got another reply from Mark. He reiterated that there's a bug, possibly several, related to searching for SG's on PRPnet.

I'm guessing that no one has ever done it before.

PRPnet testing for twins it is, with manual tests for SG's.

gd_barnes · 2023-02-05, 08:45

It looks like the server will run out of work in about a day. Shoot me the new sieve file(s) and I will load them up.

MooMoo2 · 2023-02-05, 20:04

Quote:

Originally Posted by gd_barnes

It looks like the server will run out of work in about a day. Shoot me the new sieve file(s) and I will load them up.

Sure! Here's n=1.7M: https://pixeldrain.com/u/sRnSBDmk

It's sieved to p=540T. The k=1-750T range is massive, so I've only uploaded the k=1-8.85T portion for now, which contains 2,756,047 candidates. The smaller range should be enough to last us for at least a year, though.

n=3.322M will be uploaded a bit later.

gd_barnes · 2023-02-06, 01:16

Thanks. I have loaded the new file for k<100G into port 12000 to start with. This amounted to 31225 lines.

At its current rate, n=480K-500K will finish at ~4-5 AM server time (CST U.S.), ~10-11 AM GMT, tomorrow morning.

I was able to do this because the server hands out tests by k-value and the new file had nothing for k<1M where are current tests are at. So it will seamlessly start handing out the larger tests.

This is only a "starter" load for now so the server doesn't dry out. After everything is done for n=480K-500K and I've processed and removed it from the server, we can then decide how large of pieces to load into the server at a time. k=1T ranges of ~312000 lines make sense but we can decide that based on how popular the effort is.

I need to think about the timing of how and when to snap-shot and remove the stats and how to handle straggling tests for n=480K-500K when n=1.7M work is being handed out. I will be available when n=480K-500K finishes. I may decide to stop the server for a brief period.

The Carnivore · 2023-02-06, 01:21

Just checked the stats page and saw that all n's above n=493K have been wiped and that no tests for n>493K, k=998655 were handed out. What gives?

gd_barnes · 2023-02-06, 01:25

Quote:

Originally Posted by The Carnivore

Just checked the stats page and saw that all n's above n=493K have been wiped and that no tests for n>493K, k=998655 were handed out. What gives?

Because of the nature of our current effort, the page takes ~30 mins or so to refresh itself after new candidates are loaded. I just loaded a few new n=1.7M candidates.

Nothing is lost. If you keep hitting refresh, you'll see the white lines slowly come back as they work their way down the page.

This has also happened whenever we've loaded new n=480K-500K candidates.

With the n=1.7M effort, there will just be one line so this won't appear so strange.

The Carnivore · 2023-02-06, 01:27

Quote:

Originally Posted by gd_barnes

I need to think about the timing of how and when to snap-shot and remove the stats and how to handle straggling tests for n=480K-500K when n=1.7M work is being handed out.

Perhaps the expiration for the n=480K-500K tests can be changed from 24 hours to 1 hour.

The Carnivore · 2023-02-06, 01:30

Quote:

Originally Posted by gd_barnes

Because of the nature of our current effort, the page takes ~30 mins or so to refresh itself after new candidates are loaded. I just loaded a few new n=1.7M candidates.

Nothing is lost. If you keep hitting refresh, you'll see the white lines slowly come back as they work their way down the page.

This has also happened whenever we've loaded new n=480K-500K candidates.

With the n=1.7M effort, there will just be one line so this won't appear so strange.

Understood, thanks for the explanation.

gd_barnes · 2023-02-06, 01:32

Quote:

Originally Posted by The Carnivore

Perhaps the expiration for the n=480K-500K tests can be changed from 24 hours to 1 hour.

Good plan. I've done things like that in the past to more quickly finish ranges.

Announcement: After the final tests have been handed out for n=480K-500K, I will temporarily change the expiration to 1 hour. If you have machines that hold candidates over night, those tests may expire. I will do what I can to avoid any n=1.7M candidates from expiring before 24 hours.

kruoli · 2023-02-06, 11:59

It looks like all 480k-500k candidates are now done. (Only two are reserved by one of my machines, but they are already done and are waiting for being submitted). The new candidates are being handed out successfully.

Does someone create a Web Archive of the found primes on some way or another?

gd_barnes · 2023-02-06, 12:25

Quote:

Originally Posted by kruoli

It looks like all 480k-500k candidates are now done. (Only two are reserved by one of my machines, but they are already done and are waiting for being submitted). The new candidates are being handed out successfully.

Does someone create a Web Archive of the found primes on some way or another?

There is a link to them on the TPS status page here:
http://www.noprimeleftbehind.net/tps/

I think there were also some other primes on our old LLRnet server. MooMoo kept a copy of everything.

2023-02-06, 01:21	#38
The Carnivore Jun 2010 2³·3·11 Posts	Just checked the stats page and saw that all n's above n=493K have been wiped and that no tests for n>493K, k=998655 were handed out. What gives? Attached Thumbnails

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
A question about twin primes and twin practical numbers	sweety439	sweety439	1	2022-04-23 14:32
find very easy twin prime in the infamy twin primes	hal1se	Miscellaneous Math	13	2018-11-05 16:34
pie chart: LL attempts	sixblueboxes	PrimeNet	8	2014-04-18 14:46
Next steps for TPS after Primegrid's record twin discovery	axn	Twin Prime Search	7	2011-12-31 07:04
LL-D attempts and successes	Christenson	Information & Answers	1	2011-02-03 05:25

2023-01-29, 22:02	#34
gd_barnes "Gary" May 2007 Overland Park, KS 2×7×13×67 Posts	I got another reply from Mark. He reiterated that there's a bug, possibly several, related to searching for SG's on PRPnet. I'm guessing that no one has ever done it before. PRPnet testing for twins it is, with manual tests for SG's.

2023-02-05, 08:45	#35
gd_barnes "Gary" May 2007 Overland Park, KS 2FA2₁₆ Posts	It looks like the server will run out of work in about a day. Shoot me the new sieve file(s) and I will load them up.

2023-02-06, 01:16	#37
gd_barnes "Gary" May 2007 Overland Park, KS 2·7·13·67 Posts	Thanks. I have loaded the new file for k<100G into port 12000 to start with. This amounted to 31225 lines. At its current rate, n=480K-500K will finish at ~4-5 AM server time (CST U.S.), ~10-11 AM GMT, tomorrow morning. I was able to do this because the server hands out tests by k-value and the new file had nothing for k<1M where are current tests are at. So it will seamlessly start handing out the larger tests. This is only a "starter" load for now so the server doesn't dry out. After everything is done for n=480K-500K and I've processed and removed it from the server, we can then decide how large of pieces to load into the server at a time. k=1T ranges of ~312000 lines make sense but we can decide that based on how popular the effort is. I need to think about the timing of how and when to snap-shot and remove the stats and how to handle straggling tests for n=480K-500K when n=1.7M work is being handed out. I will be available when n=480K-500K finishes. I may decide to stop the server for a brief period.

2023-02-06, 11:59	#43
kruoli "Oliver" Sep 2017 Porta Westfalica, DE 1,559 Posts	It looks like all 480k-500k candidates are now done. (Only two are reserved by one of my machines, but they are already done and are waiting for being submitted). The new candidates are being handed out successfully. Does someone create a Web Archive of the found primes on some way or another?