mersenneforum.org Team Sieve #28: c162 from 4788:2714
 Register FAQ Search Today's Posts Mark Forums Read

2011-10-21, 06:58   #23
henryzz
Just call me Henry

"David"
Sep 2007
Cambridge (GMT)

26·89 Posts

Quote:
 Originally Posted by Batalov Welcome to the club; don't feel bad. prime95 stable $\not=$ msieve stable, very true. Parallel algebra is picky in regards with the bus and NB stability. It is not as bad as it was when you would wait until "progress 101%" to face-palm and start over. The not-so-new-now orthogonality check is a much kinder solution. Good luck!
I have reached the sqrts without a problem. Reducing the memory speed didn't improve things. Should I be altering the settings for the nb(not certain I have access on my rubbish motherboard)? It is a Q6600 in case you haven't realized.
Fortunately Prime95 stability is what I am worried about not msieve stability.

Factors should hopefully be available when I come back from uni today.

2011-10-21, 11:44   #24
henryzz
Just call me Henry

"David"
Sep 2007
Cambridge (GMT)

26·89 Posts

Quote:
 Originally Posted by henryzz I have reached the sqrts without a problem. Reducing the memory speed didn't improve things. Should I be altering the settings for the nb(not certain I have access on my rubbish motherboard)? It is a Q6600 in case you haven't realized. Fortunately Prime95 stability is what I am worried about not msieve stability. Factors should hopefully be available when I come back from uni today.
I have the factors:
prp58 factor: 6154716799577446885023090856129309198086450129268256055737
prp104 factor: 27073317568454671115447910049600122939037137794629195952679986797162648298087899683780597042901407700459

The filtering and postprocessing log is attached. After the first error cost me the first 25% I stopped and restarted creating a checkpoint 2-3 times a day(~8%). Only the one of the restarts had an error on the second attempt and it had somehow created a checkpoint just before.
Attached Files
 alq_4788_2714.txt (50.3 KB, 76 views)

 2011-10-21, 16:53 #25 chris2be8     Sep 2009 34·23 Posts A few thoughts: msieve 1.43 isn't the latest version. Would the latest SVN do better? 20043267 duplicates in 96050055 relations seems a lot. Would sieving with 15e have produced fewer duplicates (assuming it produces more relations per special Q)? Chris K
 2011-10-22, 06:16 #26 debrouxl     Sep 2009 977 Posts msieve 1.49 saves multiple checkpoints for the LA, which can save days of work if a corruption happens near the end of the job. ~20M duplicates in ~96M raw relations is a bit more than 20% of duplicates. Among the RSALS jobs, there were slightly worse ratios with 14e
 2011-10-22, 06:22 #27 frmky     Jul 2003 So Cal 111111010002 Posts 20% duplicates isn't bad at all. Using 15e on larger problems, we usually get 25-27% duplicates with 2 large primes on each side, and 35% duplicates with 3 large primes on one side.
 2011-10-25, 08:51 #28 henryzz Just call me Henry     "David" Sep 2007 Cambridge (GMT) 26×89 Posts I believe I have solved the stability problems I experienced during this run. I romoved my 2x1Gb memory sticks leaving me with 2x2Gb and my pc now passes the middle stress test setting in prime95. It passed the small fft setting before I did anything. All my testing is within the small fft boundary so I don't have to worry about results. It was only about 2-3 errors in 24 hours at most anyway. I will do a postprocessing job at somepoint soon to prove I can do it without problems.

 Similar Threads Thread Thread Starter Forum Replies Last Post jrk Aliquot Sequences 55 2013-02-07 01:39 schickel Aliquot Sequences 64 2011-02-19 02:28 jrk Aliquot Sequences 31 2010-12-30 21:33 schickel Aliquot Sequences 153 2010-11-09 07:39 10metreh Aliquot Sequences 77 2009-05-27 20:39

All times are UTC. The time now is 06:41.

Wed Aug 12 06:41:27 UTC 2020 up 26 days, 2:28, 1 user, load averages: 1.03, 1.36, 1.47