Fast Breeding (guru management)

[swellman]

Anybody got a factoring job for 14e? I am considering GC_3_535, a SNFS 257 job and one of the Ten Most Wanted for the GCW project.

It should rightfully be run on 15e, but test sieving shows good results on 14e (and 14e is very hungry).

Any objections if I queue GC_3_535 on 14e once I get verification of the reservation from Paul and Sam?

[RichD]

Nothing from me for a couple days. Go for it...

[RichD]

QUEUED AS 100__427

C194 from the OPN t800 file.

[ a.k.a. Phi_5(Phi_3(Phi_3(Phi_3(Phi_3(Phi_3(Phi_7(2851)/18542517953)/2311)/3/13/5737/p13)/3/63793/25419181/55312129)/3/277)/3/523)/162091/444461 ]

P52^5-1

n: 13911732559035736832434060504081252614418620766601822872530965385868719145626982826791116903590748248963452600916277943730431252067716188708146483830579087465623937705096094313266738937571365031
# 1000560710422099960571670979677623210269290041070427^5-1, difficulty: 204.00, skewness: 1.00, alpha: 1.45
# cost: 1.99678e+17, est. time: 95.08 GHz days (not accurate yet!)
skew: 1.000
c4: 1
c3: 1
c2: 1
c1: 1
c0: 1
Y1: -1
Y0: 1000560710422099960571670979677623210269290041070427
type: snfs
rlim: 16700000
alim: 16700000
lpbr: 29
lpba: 29
mfbr: 58
mfba: 58
rlambda: 2.5
alambda: 2.5

Trial sieving 5K blocks.

  Q  Yield
 10M  9566
 20M 11142
 30M 10302
 40M  9579
 42M  9248

[RichD]

QUEUED AS 19531_59m1

C219 from the OPN t550 file.

n: 287018215008985551660511535628614538194044207507659437134898068193842795789963836763626109269023880558531848923349941210319133222229962467454957044999743408664002003264153104567564975538169594778284972238419193134482747
# 19531^59-1, difficulty: 257.44, skewness: 5.19, alpha: 0.00
# cost: 1.45874e+19, est. time: 6946.39 GHz days (not accurate yet!)
skew: 5.189
c6: 1
c0: -19531
Y1: -1
Y0: 8076901753252440873102885328904621929075801
type: snfs
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.7
alambda: 2.7

Trial sieving 5K blocks.

  Q  Yield
 20M 13150
 60M 10807
100M 10466
150M  8789

[RichD]

My last two jobs apparently were inadvertently tested with the 15e siever. 100__427 might be salvageable by increasing the Q until we get enough relations. 19531_59m1 will never get enough no matter how far Q is extended. So it needs to be removed from the queue.

From the forensics I suspect I may have tried to trick a script into using the 15e instead of 14e for some testing by renaming 15e to 14e, I also have a 14ea siever which might be the original 14e. I normally don't use this machine for testing but it was a bit idle ATM.

I also have two additional jobs ready to go but they will need to be re-tested with the correct siever.

[swellman]

Originally Posted by RichD

My last two jobs apparently were inadvertently tested with the 15e siever. 100__427 might be salvageable by increasing the Q until we get enough relations. 19531_59m1 will never get enough no matter how far Q is extended. So it needs to be removed from the queue.

From the forensics I suspect I may have tried to trick a script into using the 15e instead of 14e for some testing by renaming 15e to 14e, I also have a 14ea siever which might be the original 14e. I normally don't use this machine for testing but it was a bit idle ATM.

I also have two additional jobs ready to go but they will need to be re-tested with the correct siever.

Only Greg can kill a job but I can reach out to him if you think 19531_59m1 cannot finish as submitted. But it may come close - do you want to add more Q and see if a matrix can be built?

I will add more Q to 100__427.

Eta: NFS@Home suggests Qmax = 202M. I think it’s more like 220M but that’s not a ridiculous value for Q.

[RichD]

Originally Posted by swellman

Only Greg can kill a job but I can reach out to him if you think 19531_59m1 cannot finish as submitted. But it may come close - do you want to add more Q and see if a matrix can be built?

Since there is nothing waiting in the queue let it run, otherwise what has already been done would be wasted. I don't know how far to extend the Q before it peters out.

I will do the post-processing and if more is needed I will use VBCurtis suggestion and run low Q's with 15e to hopefully finish it.

[VBCurtis]

Originally Posted by RichD

I will do the post-processing and if more is needed I will use VBCurtis suggestion and run low Q's with 15e to hopefully finish it.

I've been experimenting with using CADO and a larger siever on low Q's to help finish jobs like this. You might choose to run Q from e.g. 2-20M on CADO with I=15; if you do, expect a *bunch* of relations.

[RichD]

Originally Posted by VBCurtis

I've been experimenting with using CADO and a larger siever on low Q's to help finish jobs like this. You might choose to run Q from e.g. 2-20M on CADO with I=15; if you do, expect a *bunch* of relations.

Is the relation file the same format as GGNFS?

[VBCurtis]

Yep! Each workunit is stored as its own .gz file; zcat'ing the lot into a single .dat file can be fed into remdups or msieve without any alteration.

The directory is "{jobname}.upload" inside the work directory of the factorization.

Once in there, zcat *.gz > relations.dat

relations.dat can then be cat'ed with your nfs@home download.

[RichD]

QUEUED AS 316950847_29m1

C161 from the OPN t600 file.

n: 51814107693707719941473139963487636114544745623643230615744171243250904745843234383981869668992127527676517915442567054703161432272120999750112834346368573174589
# 316950847^29-1
# expecting poly E from 1.14e-12 to > 1.31e-12
lss: 0
Y0: -10220890163452387005521020897678
Y1: 4188446172089267
c0: 851968152979185641676454807362709505595
c1: 1070535904148754871512115443325548
c2: 80834844168441856949303179
c3: -60867414129584661032
c4: -625916300066
c5: 464520
skew: 6533178.50
# skew 6533178.50, size 1.155e-15, alpha -7.058, combined = 1.234e-12 rroots = 5
type: gnfs
rlim: 67000000
alim: 67000000
lpbr: 30
lpba: 30
mfbr: 60
mfba: 60
rlambda: 2.6
alambda: 2.6

Trial sieving 5K blocks.

  Q  Yield
 12M  8933
 20M 10078
 50M  9501
 80M  6859
 90M  8125