2019 14e post processing reservations and results

[RichD]

Originally Posted by jyb

Given the high duplicate rate,

I wouldn't think a GNFS job would have a high duplicate rate but I really don't know. I'm just starting to dabble with 32-bit jobs.

[jyb]

Originally Posted by RichD

I wouldn't think a GNFS job would have a high duplicate rate but I really don't know. I'm just starting to dabble with 32-bit jobs.

I can't really comment on the theory here. But whether or not there should be a lot of duplicates, there unquestionably are a lot of duplicates.

I have noticed in the past that the relations produced by NFS@Home sieving seem to have way more bad lines and duplicates than I get when I do jobs myself. Perhaps there are some benign explanations for this, but I have heard it suggested that it may be caused by various forms of "cheating" when people returns relations to the server.

In any case, when I submit jobs I tend to suggest special-q ranges that produce way more relations than I would try for if I were doing the sieving myself, as a way of getting around this problem.

[RichD]

Originally Posted by jyb

I can't really comment on the theory here. But whether or not there should be a lot of duplicates, there unquestionably are a lot of duplicates.

I have noticed in the past that the relations produced by NFS@Home sieving seem to have way more bad lines and duplicates than I get when I do jobs myself. Perhaps there are some benign explanations for this, but I have heard it suggested that it may be caused by various forms of "cheating" when people returns relations to the server.

In any case, when I submit jobs I tend to suggest special-q ranges that produce way more relations than I would try for if I were doing the sieving myself, as a way of getting around this problem.

Ah, yes, I am starting to notice that. The bigger the job the more garbage it collects. The longer it is in the queue the more likelihood of mistreatment.

31-bit jobs need more than the expected.

BTW, my 3372531985651_19m1 job built a matrix at TD=132 and should complete on or about Friday, 5 July if all goes well.

[RichD]

Taking C180_159978_10330.

[richs]

Originally Posted by richs

Reserving C182_M17_k98

p56 factor: 76895458070627070525007761656219867594868038627047528439
p126 factor: 194670777427975176331993953351170360734240081917106572346085377818707289417171889284456923645153422980830974783021883407250207

Approximately 25.7 hours on 2 threads Core i3-2310M with 4 GB memory for a 3.45M matrix at TD = 100.

Full log attached and redacted log (due to numerous relation errors from some cheater) at https://pastebin.com/xfaHqcLe

Factors added to FDB.

[jyb]

38160408950501_17m1 needs more sieving. Over 35% of the relations are duplicates, and msieve doesn't get very far in filtering before it complains. I'd probably add another 40M.

Done.

[fivemack]

Taking 10m9_251

[richs]

Taking 10p9_263

[jyb]

Taking C164_3366_2191.

[Dylan14]

C166_M19_k74 is complete:

p54 factor: 387184084614656438987154506853071394996491008532723507
p112 factor: 7858374697721506817025694875934862172411858587345378675715373341089477652165751964689422482397011529641725140733

TD = 122 built a 4.56 M matrix (not TD 125) which took 8h 37m to solve. Log posted at https://pastebin.com/jN9TzLfp and https://github.com/Dylan1496/nfs-at-...66_M19_k74.log.
Reserving 5p3_1425M for post processing.

[jyb]

p54 factor: 308312007300944575027791293633536467795756007147545161
p110 factor: 36997115764656058946624603757199691170627538192760153675664831011355819056889320458078857874833318561030674629

https://pastebin.com/N2qMrvy9