Linear algebra reservations, progress and results

[fivemack]

Sun Nov 20 20:07:11 2016  p84 factor: 382287215827374516490913335371311113681067958598268402528490562590751491450318848599
Sun Nov 20 20:07:11 2016  p157 factor: 3400940389968403667731539737448696123231575042827382632730660962351212715477766287828544695265357156149504130573165768485381955913367322088713721160358336633

70.3 hours for 11.4M density-152 matrix on 6 threads i7-4930K. Log attached and at http://pastebin.com/Cba9U0a6

[pinhodecarlos]

Originally Posted by pinhodecarlos

Taking 8-7_305 (14e).

Can't manage to build a matrix with TD set to 140 and 130.

[fivemack]

Sat Nov 19 00:32:32 2016  p108 factor: 716905281051458361496599418461912038526878202032358002696049777126802136327392049986279278218827331523325193
Sat Nov 19 00:32:32 2016  p115 factor: 1985539808000448864336423843924848069332363172535981202002959354687318573297260097054958575980797727863937374396257

About 376 hours for 26.4M density-110 matrix on 7 threads E5-2650v2 (two intervening reboots)

Log attached and at http://pastebin.com/BhQAzWsm

[pinhodecarlos]

Originally Posted by pinhodecarlos

Can't manage to build a matrix with TD set to 140 and 130.

Thank you Jon for the extra sieving. Trying once again, I'll came back within 3 hours to share if LA is underway.

[jyb]

Originally Posted by pinhodecarlos

Thank you Jon for the extra sieving. Trying once again, I'll came back within 3 hours to share if LA is underway.

Actually, I think that doing more sieving was the wrong choice. The reason the matrix wouldn't build was that there were too many relations. If you cut it down (I tried filter_maxrels=250000000), then it manages to finish the filtering just fine with td=130.

[pinhodecarlos]

Originally Posted by jyb

Actually, I think that doing more sieving was the wrong choice. The reason the matrix wouldn't build was that there were too many relations. If you cut it down (I tried filter_maxrels=250000000), then it manages to finish the filtering just fine with td=130.

But shouldn't the matrix be built even when you have oversieved SNFS220? Anyway, I'm already in bed and filtering is still going on and I can see twice message "matrix not dense enough, retrying". Think another hour and LA will be running but I won't be awake to control this.

Ps( using my iPhone to connect to UK home machine).

[jyb]

Originally Posted by pinhodecarlos

But shouldn't the matrix be built even when you have oversieved SNFS220? Anyway, I'm already in bed and filtering is still going on and I can see twice message "matrix not dense enough, retrying". Think another hour and LA will be running but I won't be awake to control this.

Ps( using my iPhone to connect to UK home machine).

I don't know the details (I'm sure jasonp could enlighten us), but I know that there are times when too many relations will result in just the situation you're seeing. The solution is to cut down the number of relations a bit.

In any case, I ran the filtering with 250M relations, and it built a matrix without any trouble.

[pinhodecarlos]

Originally Posted by jyb

I don't know the details (I'm sure jasonp could enlighten us), but I know that there are times when too many relations will result in just the situation you're seeing. The solution is to cut down the number of relations a bit.

In any case, I ran the filtering with 250M relations, and it built a matrix without any trouble.

Just stayed talking with my wife whilst awaiting for the LA to start and it's underway with an ETA of 71 hours. Almost midnight here so now I'm going to sleep.

[unconnected]

Reserving 3366:i2156.

[jyb]

Reserving 127^93+93^127 cofactor (15e).

[unconnected]

Reserving C229_148_70.