Linear algebra reservations, progress and results

[richs]

I ran msieve on C237_12161_59 (log attached) and when the full merge process is reached, msieve goes dormant. By dormant, task scheduler shows about 0.3% cpu usage with 5+ Gb memory with the amount of memory varying every refresh. With the prior number that I ran, full merge took less than 10 minutes. Perhaps not enough memory? Running Windows 10 with 8 Gb memory.

Any advice would be appreciated.

[swellman]

I've had this same thing happen and it seemed to be insufficient memory. Going to another machine with more memory solved the problem, though you could try to rerun with TD=100 or 110 just to try to build the matrix. Or sieve more relations(?).

When this happened to me, I let msieve run for well over a week until full merge finally completed. The time to sieve estimate was measured in years!

[fivemack]

Fri Apr 21 13:23:57 2017  p68 factor: 69850751451692983626652885615515723472011733874399458655606950777087
Fri Apr 21 13:23:57 2017  p135 factor: 653748673101791536278347326477209219516452170215871376896498224735082845923886366677483702388290438325532576068540665597626205444877929

91.8 hours on 6 threads i7/4930K for a 13.85M matrix at density 108 (120 didn't work).

Log attached and at https://pastebin.com/nLdwpBdB

I had to run the linear algebra twice - on the first attempt it solved the matrix but every relation produced an 'error: relation 166785513 corrupt' or 'error: relation 185807891 corrupt' message at the square root phase, except one that produced the error 'algebraic side is not a square'. I did delete lines 166785513 and 185807891 from the file before repeating, but neither had anything visibly odd about it.

[richs]

Originally Posted by richs

I ran msieve on C237_12161_59 (log attached) and when the full merge process is reached, msieve goes dormant. By dormant, task scheduler shows about 0.3% cpu usage with 5+ Gb memory with the amount of memory varying every refresh. With the prior number that I ran, full merge took less than 10 minutes. Perhaps not enough memory? Running Windows 10 with 8 Gb memory.

Any advice would be appreciated.

The good news is that booting Windows 10 in safe mode to a command prompt and setting target density = 70 allowed the full merge to proceed. The bad news is that the linear algebra ETA is 18+ days.

[RichD]

I'll take C206_149xx041_13 next.

I will be relocating this box so the post-processing might be piecemeal.

[fivemack]

Tue Apr 18 19:55:53 2017  p87 factor: 298893797632505694369365366807674889411602696800624949866982229091912376171341334516301
Tue Apr 18 19:55:53 2017  p115 factor: 4095794635032719428105910461099175067710721384082052133754787896001385421174172773504078872241406037617843306535607

(but I was on a training course last week and only got to see this computer again today)

271.1 hours for 22.8M density-120 matrix on 7 cores E5-2650v2.

Log attached and at https://pastebin.com/Rze8g393

[fivemack]

Taking C187_142_67

[swellman]

p54 factor: 159790103673140365135029060471459644167485842180777189
p197 factor: 31986405876337473667915871421471627616138122756251708646798873692125179282754424937563555845773902840234802484347893904745362282712769989163180104828654450774930325076753204879388780056993644766371

[unconnected]

Please report it to near-repdigit project here: http://stdkmd.com/nrr/c.cgi?q=51119_249

[swellman]

Originally Posted by unconnected

Please report it to near-repdigit project here: http://stdkmd.com/nrr/c.cgi?q=51119_249

Done. Also reported it to factirdb earlier.

[swellman]

Originally Posted by fivemack

Taking C187_142_67

I had reserved this a few days ago, but I haven't downloaded it yet so I'll just grab another.

Reserving C207_128_91.