mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Reply
 
Thread Tools
Old 2010-01-05, 07:31   #34
Brucifer
 
Brucifer's Avatar
 
Dec 2005

313 Posts
Default

2900-3000-50k-250k is completed and the results have been emailed to Max.

Last fiddled with by Brucifer on 2010-01-05 at 07:31
Brucifer is offline   Reply With Quote
Old 2010-01-05, 09:09   #35
kar_bon
 
kar_bon's Avatar
 
Mar 2006
Germany

22·691 Posts
Default

now it's my time to fill in the whole range upto n=250k, i think.

this could take a bit time so the next range n>250k should be started after that work is done, please!
kar_bon is offline   Reply With Quote
Old 2010-01-29, 05:39   #36
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3·2,083 Posts
Default

I noticed that this thread hasn't had its status table updated for a while despite the fact all the ranges shown within as "in progress" are now long complete. Gary, I know I've processed all of Bruce's results and sent them to you; did you get the processed results from Karsten (I think he's doing those, right?) from the LLRnet portion of the drive (k<2400)?

I held off on marking Bruce's range as complete since Ian needs to mark down all the primes for it first. Ian, I think I recall you mentioned somewhere that you were waiting for Gary to move out the primes in the table to a linked file before you could add more; if you need to, you could just put them in a text file and attach it to a post here. Then Gary could upload them and put the links in the table.
mdettweiler is offline   Reply With Quote
Old 2010-01-29, 12:20   #37
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

235518 Posts
Default

Quote:
Originally Posted by mdettweiler View Post
I noticed that this thread hasn't had its status table updated for a while despite the fact all the ranges shown within as "in progress" are now long complete. Gary, I know I've processed all of Bruce's results and sent them to you; did you get the processed results from Karsten (I think he's doing those, right?) from the LLRnet portion of the drive (k<2400)?

I held off on marking Bruce's range as complete since Ian needs to mark down all the primes for it first. Ian, I think I recall you mentioned somewhere that you were waiting for Gary to move out the primes in the table to a linked file before you could add more; if you need to, you could just put them in a text file and attach it to a post here. Then Gary could upload them and put the links in the table.
Oh, I moved out the primes long before Bruce commenced with the manual processing of the range. Ian and I just need to follow up on listing all of the primes. It's an arduous task that I just haven't had time for yet. What usually has been happening is that he would list as many as he could until it fills up, then I would move them off. But for tons of them that came in so fast like that, it makes more sense for me to just format them first into file links to begin with and maybe have him list the last k=100 or so of them. In other words, it makes no sense for him to list them, then have me copy them off, then him list them, etc.

So it's on my shoulders, not his, to put them in file links at this point. It's a bit of an effort because I have to check whether they are confirmed or new, just like Ian does when he lists them in the 1st post here. But what is a pain to get started on it this time around is that I need to wade through Bruce's primes to see which one's Lennart's already found. I think I'll show them both as finding them.

So as you can see, it's no easy task. Tell you what...Monday is my admin day when I'm out of town...good day to catch up on paperwork and stuff. I'll write it on my task list for Monday to spend an hour checking what is new/confirmed, if Bruce/Lennart both found the same prime, etc. and creating file links. In an hour, I'm thinking I can finish the most difficult part, which is doing k=2400-2600. Even though we already have primes well above that listed, Bruce started manually at k=2400 because of the server problems that we had so I'll likely need to show both of them finding quite a few primes. Another day I'll get to k=2600-3000, which should take less time for twice the size since it's a clean manual run. I guess that's why I've procrastinated it a little...I know k=2400-2600 is going to be a pain administratively.

Karsten, please hold off on listing in k=2400-3000 for n=50K-250K primes found in this effort on your pages until after next Monday. That way, I can tell if they were confirmed or found new.

Regardless, Max, if you have time, I'd like to see LLRnet port 7000 get loaded up with k=2000-2100 (or 2000-2200) for n=250K-350K any time and then let Bruce know about it. LLRnet works great for this range. We only goofed last time because of the extra header-in-the-file problem. You might give him your personal guarantee that we've closely inspected the file for any error that could cause problems and that my LLRnet port 6000 has been running great with ~60-70 cores total on it right now. Also mention that since the tests aren't so small, the load on the server will be much smaller. (n=250K takes ~25 times as long to test as n=50K)


Gary
gd_barnes is offline   Reply With Quote
Old 2010-01-29, 14:37   #38
kar_bon
 
kar_bon's Avatar
 
Mar 2006
Germany

22×691 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Karsten, please hold off on listing in k=2400-3000 for n=50K-250K primes found in this effort on your pages until after next Monday. That way, I can tell if they were confirmed or found new.
As you can see on rieselprime for Drive 12A, I'm at k=2177 with including primes in the 2000-4000 summary page so far.
It's indeed very time-consuming to check for confirmed primes but i found some problems so far:
There're some primes not listed yet in the summary although the test-range was higher. Here are the issues so far:

2055: done to n=400k (S.Tjung) but missed n=216369 -> max-n=250k
2085: done to n=400k (S.Tjung) but missed n=193924 -> max-n=250k
2171: done to n=1.0M (Beyond?) but missed n=101974 -> max-n=250k
2175: done to n=400k (S.Tjung) but missed n=183243,224093 -> max-n=250k

I've therefore set the test-range down to n=250k.

More work will be done slowly the next weeks!
Don't worry, Gary, you got enough time.

Last fiddled with by kar_bon on 2010-01-29 at 14:39
kar_bon is offline   Reply With Quote
Old 2010-01-29, 17:34   #39
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Oh, I moved out the primes long before Bruce commenced with the manual processing of the range. Ian and I just need to follow up on listing all of the primes. It's an arduous task that I just haven't had time for yet. What usually has been happening is that he would list as many as he could until it fills up, then I would move them off. But for tons of them that came in so fast like that, it makes more sense for me to just format them first into file links to begin with and maybe have him list the last k=100 or so of them. In other words, it makes no sense for him to list them, then have me copy them off, then him list them, etc.

So it's on my shoulders, not his, to put them in file links at this point. It's a bit of an effort because I have to check whether they are confirmed or new, just like Ian does when he lists them in the 1st post here. But what is a pain to get started on it this time around is that I need to wade through Bruce's primes to see which one's Lennart's already found. I think I'll show them both as finding them.

So as you can see, it's no easy task. Tell you what...Monday is my admin day when I'm out of town...good day to catch up on paperwork and stuff. I'll write it on my task list for Monday to spend an hour checking what is new/confirmed, if Bruce/Lennart both found the same prime, etc. and creating file links. In an hour, I'm thinking I can finish the most difficult part, which is doing k=2400-2600. Even though we already have primes well above that listed, Bruce started manually at k=2400 because of the server problems that we had so I'll likely need to show both of them finding quite a few primes. Another day I'll get to k=2600-3000, which should take less time for twice the size since it's a clean manual run. I guess that's why I've procrastinated it a little...I know k=2400-2600 is going to be a pain administratively.
Well, what I had in mind was that possibly Ian could take the primes and put them into flat text files, which he would attach to a forum post and you could upload to the web. Then you could just plop the file links here. (We wouldn't even necessarily need to list the last k=100 or so of them...there's so many of them that I imagine even just the file links will fill up the code box somewhat.) This is, of course, assuming that Ian doesn't mind.

As for k=2400-2600: that might be a tad tricky. You see, the results files from the PRPnet portion of that never got copied off due to the database mess that needs to be cleaned up; they're still sitting in G7000's folder on humpford's desktop. I can send the file to you if you'd like.

Quote:
Karsten, please hold off on listing in k=2400-3000 for n=50K-250K primes found in this effort on your pages until after next Monday. That way, I can tell if they were confirmed or found new.

Regardless, Max, if you have time, I'd like to see LLRnet port 7000 get loaded up with k=2000-2100 (or 2000-2200) for n=250K-350K any time and then let Bruce know about it. LLRnet works great for this range. We only goofed last time because of the extra header-in-the-file problem. You might give him your personal guarantee that we've closely inspected the file for any error that could cause problems and that my LLRnet port 6000 has been running great with ~60-70 cores total on it right now. Also mention that since the tests aren't so small, the load on the server will be much smaller. (n=250K takes ~25 times as long to test as n=50K)
About that, I'm still waiting for the database to be cleaned out from the old mess. I'll have to send Dave an email about that today.
mdettweiler is offline   Reply With Quote
Old 2010-01-30, 00:43   #40
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

32·19·59 Posts
Default

Quote:
Originally Posted by kar_bon View Post
As you can see on rieselprime for Drive 12A, I'm at k=2177 with including primes in the 2000-4000 summary page so far.
It's indeed very time-consuming to check for confirmed primes but i found some problems so far:
There're some primes not listed yet in the summary although the test-range was higher. Here are the issues so far:

2055: done to n=400k (S.Tjung) but missed n=216369 -> max-n=250k
2085: done to n=400k (S.Tjung) but missed n=193924 -> max-n=250k
2171: done to n=1.0M (Beyond?) but missed n=101974 -> max-n=250k
2175: done to n=400k (S.Tjung) but missed n=183243,224093 -> max-n=250k

I've therefore set the test-range down to n=250k.

More work will be done slowly the next weeks!
Don't worry, Gary, you got enough time.

That's why NPLB includes ALL k's in our drives (unless reserved and we would "run over" existing reservations). Way to go all! We found primes missed by others! Even the non-top-5000 primes are important.

On the 5 or so k's excluded from the 11th drive, we'll (mainly me but also whomever else might want to help) will be double checking those ~5 k's for n=425K-600K close to the time that the 11th drive is nearing n=600K. I'm shooting for mid-year on it getting there. Late in 2008, such a double check netted a missing top-5000 prime!

k=2400-2600 is really messy so I'll work on sorting it out and getting a file link for it on Monday. It's high on my task list now.

Max, good idea on just doing file links instead of listing the last 100k or so in the post. That's a bit of a nuisance. Yeah, I/we will just do file links for all of the primes...well, at least for all n=50K-250K primes. The file links are small and use little posting space. With all n=50K-250K primes in links, we'll likely be able to show all or a large portion of n=250K-350K primes as line items in the post. That's kind of a personal preference on my part. :-)

Max, one more thing. All of Lennart's k=2400-2600 primes are already shown in the 1st post here. I'll just take the primes file that you already sent me from Bruce (or that he sent me?) and add those to it. For primes that they both found, I'll just show both their names as having found them, with whomever found them first listed first. In other words, you don't need to do anything for me to be able to do that. That's my Monday task. :-)


Gary

Last fiddled with by gd_barnes on 2010-01-30 at 00:49
gd_barnes is offline   Reply With Quote
Old 2010-02-02, 04:33   #41
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

32·19·59 Posts
Default

Unfortunately I won't be able to get all of the primes brought together today for k=2400-2600. What I'm running into is that Bruce's list of primes from his manual LLRnet run is missing at least a couple of primes from the original run that we did on a public PRPnet/LLRnet server. I don't know if that's because he missed them or we have extras in the original run. Also, of course the original run here is missing many primes vs. Bruce's list, even in the middle of the run due to problems with the public server. That's not a concern but Bruce not having primes found by the public server in his list is a problem.

Fortunately Bruce was kind enough to send me/Max the immense amount of results from the range so I should be able to isolate the problems more easily.

This will take several hours to get it all together, matched up, proper credit given, and check the new/confirmed status. I'm about 1/3rd of the way into it and realized that it will take much longer than I originally expected. I'll post more details about the above problem late Tuesday.

I'll work on it more on Tuesday but can't guarantee completion depending on issues that I run into. It is my hope to at least have all of the k=2400-2600 primes listed by the end of this week. After that, k=2600-3000 should go much more quickly.


Gary

Last fiddled with by gd_barnes on 2010-02-02 at 04:41
gd_barnes is offline   Reply With Quote
Old 2010-02-02, 05:05   #42
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

276916 Posts
Default

There were 4 primes found by Lennart on the public server that were not found by Bruce in his manual LLRnet run that effectively was a double-check of a large part of the range. Triple-checking by me confirmed that Bruce's list is correct. Here are the culprits:

2457*2^239794-1 is not prime. LLR Res64: 55D32DAFD42FB9B4 Time : 198.913 sec.
2463*2^170838-1 is not prime. LLR Res64: EC031A0DB6C2DCF2 Time : 127.062 sec.
2481*2^185187-1 is not prime. LLR Res64: A64A60A7D5942FD4 Time : 119.079 sec.
2505*2^154470-1 is not prime. LLR Res64: 07BE4011A7E08FF5 Time : 89.002 sec.

In all 4 cases, these were close to actual primes that were found by both Bruce and Lennart. They are:
2457*2^239814-1 (n=20 difference from above "composite prime")
2463*2^170842-1 (n=4 difference)
2481*2^185218-1 (n=31 difference)
2505*2^154475-1 (n=5 difference)

That is exactly what happened to me on 2 occassions when I had a laptop with bad memory. It would find a "composite prime" very close to an actual prime. I'll send Lennart a PM referencing this post. I have now deleted the erroneous primes from the 1st post here.

That's a start on the cleanup of k=2400-2600 here...still much work to do on it later Tuesday.


Gary

Last fiddled with by gd_barnes on 2010-02-02 at 05:49
gd_barnes is offline   Reply With Quote
Old 2010-02-02, 05:33   #43
Lennart
 
Lennart's Avatar
 
"Lennart"
Jun 2007

25·5·7 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
There were 4 primes found by Lennart on the public server that were not found by Bruce in his manual LLRnet run that effectively was a double-check of a large part of the range. Triple-checking by me confirmed that Bruce's list is correct. Here are the culprits:

2457*2^239794-1 is not prime. LLR Res64: 55D32DAFD42FB9B4 Time : 198.913 sec.
2463*2^170838-1 is not prime. LLR Res64: EC031A0DB6C2DCF2 Time : 127.062 sec.
2481*2^185187-1 is not prime. LLR Res64: A64A60A7D5942FD4 Time : 119.079 sec.
2505*2^154470-1 is not prime. LLR Res64: 07BE4011A7E08FF5 Time : 89.002 sec.

In all 4 cases, these were close to actual primes that were found by both Bruce and Lennart. They are:
2457*2^239814-1 (n=20 difference from above "composite prime")
2463*2^170842-1 (n=4 difference)
2481*2^185218-1 (n=31 difference)
2505*2^154475-1 (n=5 difference)

That is exactly what happened to me on 2 occassions when I had a laptop with bad memory. It would fine a "composite prime" very close to an actual prime. I'll send Lennart a PM referencing this post. I have now deleted the erroneous primes from the 1st post here.

That's a start on the cleanup of k=2400-2600 here...still much work to do on it later Tuesday.


Gary

I can't answer the question. I need to know the computer (ClirntID) from log to see that. If there is a problem it should show up in Boinc becuse all is DC there so that computer should have many invalid wu,


But can you give me the log entrys frp server there clientid is i can mayby narrow it down.


Lennart
Lennart is offline   Reply With Quote
Old 2010-02-02, 05:39   #44
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

235518 Posts
Default

Quote:
Originally Posted by Lennart View Post
I can't answer the question. I need to know the computer (ClirntID) from log to see that. If there is a problem it should show up in Boinc becuse all is DC there so that computer should have many invalid wu,


But can you give me the log entrys frp server there clientid is i can mayby narrow it down.


Lennart
OK, I'll see what I can do. It may be Thursday or Friday after I get back from a business trip. It's not easy to access the servers remotely right now.
gd_barnes is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Team drive #10 k=1400-2000 n=500K-1M gd_barnes No Prime Left Behind 61 2013-01-30 16:08
Team drive #11 k=2000-3000 n=425K-600K gd_barnes No Prime Left Behind 42 2010-11-19 10:42
Team drive #9 k=1005-2000 n=50K-350K gd_barnes No Prime Left Behind 236 2009-06-25 10:04
Sieving drive for k=2000-3400 n=50K-1M gd_barnes No Prime Left Behind 145 2009-06-23 18:28
Team drive #8 k=1400-2000 n=350K-500K gd_barnes No Prime Left Behind 101 2009-04-08 02:11

All times are UTC. The time now is 07:58.

Wed Apr 1 07:58:08 UTC 2020 up 7 days, 5:31, 0 users, load averages: 1.77, 2.06, 1.91

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.