Team drive #12 k=2000-3000 n=50K-425K

[mdettweiler]

Originally Posted by Lennart

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

Right now we're having some issues with the server machine, so I might have a hard time pulling files off of it. I do still have the PPRnet G7000 original results sitting around on the server, so I'll try to get a hold of them as soon as possible and try to spot which machine(s) produced the culprit results.

[gd_barnes]

Originally Posted by mdettweiler

Right now we're having some issues with the server machine, so I might have a hard time pulling files off of it. I do still have the PPRnet G7000 original results sitting around on the server, so I'll try to get a hold of them as soon as possible and try to spot which machine(s) produced the culprit results.

Max, after I get back very late Thursday and get the machine rebooted, can you put it on your list to send Lennart this info. sometime on Friday? Thanks.

Lennart, something quirky in our remote setup is keeping us from viewing files on the server machine and one other machine right now. Sorry for the delay.


Gary

[gd_barnes]

After much effort, the listing of k=2400-2600 primes is complete. The counts of new and confirmed primes have been verified against the results files that Bruce sent and Karsten's rieselprime web page.

I'll do k=2600-3000 later Weds. It should be child's play compared to k=2400-2600.

Max, I've let this slide on a # of drives right now due to heavy admin duties on both of the projects right now but while we're at it here, can you please sent me "regular LLR" formatted results for k=2000-2400 here so that I can verify the prime counts for that range? Bruce (or you) have already sent me results for k=2600-3000. Generally I leave the 1st post status as "in progress" until I'm able to verify the results but made an exception here since it would look odd to have k=2400-2600 complete and k=2000-2400 as in progress.

On the other hand, it's possible that I could have missed an Email or two from way back so if you've already sent the k=2000-2400 results, just let me know the date that they were sent and I'll find them.


Thanks,
Gary

[mdettweiler]

Originally Posted by gd_barnes

After much effort, the listing of k=2400-2600 primes is complete. The counts of new and confirmed primes have been verified against the results files that Bruce sent and Karsten's rieselprime web page.

I'll do k=2600-3000 later Weds. It should be child's play compared to k=2400-2600.

Max, I've let this slide on a # of drives right now due to heavy admin duties on both of the projects right now but while we're at it here, can you please sent me "regular LLR" formatted results for k=2000-2400 here so that I can verify the prime counts for that range? Bruce (or you) have already sent me results for k=2600-3000. Generally I leave the 1st post status as "in progress" until I'm able to verify the results but made an exception here since it would look odd to have k=2400-2600 complete and k=2000-2400 as in progress.

On the other hand, it's possible that I could have missed an Email or two from way back so if you've already sent the k=2000-2400 results, just let me know the date that they were sent and I'll find them.


Thanks,
Gary

The results in LLRnet format should be in the same place as always; last I recall, Karsten was the one doing the processing of those results into LLR format and sending them to you. Karsten, do you have those handy by chance?

[mdettweiler]

Originally Posted by gd_barnes

Max, after I get back very late Thursday and get the machine rebooted, can you put it on your list to send Lennart this info. sometime on Friday? Thanks.

Lennart, something quirky in our remote setup is keeping us from viewing files on the server machine and one other machine right now. Sorry for the delay.


Gary

Okay. I may actually be able to copy off the file via the command line before you get the machine rebooted, but I'll be out of the house for much of tomorrow so I might not get to it then. (It's too late to do it tonight.) I'll do it as soon as I get the chance, though.

[gd_barnes]

Originally Posted by mdettweiler

The results in LLRnet format should be in the same place as always; last I recall, Karsten was the one doing the processing of those results into LLR format and sending them to you. Karsten, do you have those handy by chance?

Karsten sent me a PM with a link to them. Thanks for checking.

Originally Posted by mdettweiler

Okay. I may actually be able to copy off the file via the command line before you get the machine rebooted, but I'll be out of the house for much of tomorrow so I might not get to it then. (It's too late to do it tonight.) I'll do it as soon as I get the chance, though.

I won't be rebooting it until very late Thursday night; more like early Friday morning. So you should have most of Weds. and Thurs. to take care of it if you have time.


Gary

[mdettweiler]

Originally Posted by gd_barnes

I won't be rebooting it until very late Thursday night; more like early Friday morning. So you should have most of Weds. and Thurs. to take care of it if you have time.

Okay, I've got the file now and have found the false primes in there. But...the news is not good. I found three distinct machines that produced these results:

[2009-12-06 05:00:10 GMT] 2457*2^239794-1 received by Email: sm5ymt@*******.**  User: sm5ymt  Client: 6 Program: llr  Prover: llr  Residue: PRIME  
[2009-12-06 07:03:16 GMT] 2463*2^170838-1 received by Email: sm5ymt@*******.**  User: sm5ymt  Client: 96 Program: llr  Prover: llr  Residue: PRIME  
[2009-12-07 17:12:46 GMT] 2481*2^185187-1 received by Email: sm5ymt@*******.**  User: sm5ymt  Client: 150 Program: llr  Prover: llr  Residue: PRIME  
[2009-12-07 20:01:51 GMT] 2505*2^154470-1 received by Email: sm5ymt@*******.**  User: sm5ymt  Client: 96 Program: llr  Prover: llr  Residue: PRIME

On is from client "6", two from "96", and one from "150". The most reasonable conclusion would seem to be that all three are bad (or at least were at the time they produced these results--either that or Lennart's house got hit with a real whammy of cosmic rays around that time ).

Edit: I just thought of something. Lennart, do you name your clients individually per core, or per machine? If those are all cores of the same machine, then that might explain why there appeared to be three problematic clients.

[gd_barnes]

It's too clunky to work with the huge results files while out of town. My plan now is to work on the primes for k=2600-3000 sometime this upcoming weekend. I'm just glad to have the messy k=2400-2600 range done.

[gd_barnes]

All primes for this drive have now been confirmed and entered in the 1st post for n=50K-250K.

As soon as we can confirm the database cleanup for the range, we can begin on n=250K-350K. Max, what is the status of that?

[mdettweiler]

Originally Posted by gd_barnes

All primes for this drive have now been confirmed and entered in the 1st post for n=50K-250K.

As soon as we can confirm the database cleanup for the range, we can begin on n=250K-350K. Max, what is the status of that?

Still not happened yet. I'll email Dave about it.

[gd_barnes]

Max,

Now that the database has been cleaned up, can we go ahead and load k=2000-2200 for n=250K-350K into LLRnet port 7000?


Gary