Reserved for MF - Sequence 3366

[charybdis]

Now running curves at 4e8. How much ECM do we reckon this number will need?

[VBCurtis]

I use 0.31 * size for GNFS numbers, with a bit more for C195+ since the matrices are a bit of a bother. So, I'd personally be happy at around 1.5T60.

I did 1000 @15e7 and 500 @3e8 yesterday; I'll run some at 65e7 next, just a few hundred.

[VBCurtis]

New curves: 1000@65e7, 1000@85e7.

Last, trying 500@16e8.

[charybdis]

Currently on 29000@4e8. We've probably done just over a t60 in total; I'll move up to 1e9.

I presume this will end up in the NFS@home 15e queue if ECM doesn't crack it. Probably a good idea to start the polynomial search soon?

[VBCurtis]

Agreed, I'll start poly select Monday afternoon after this last run of curves completes.

I expect to do some local sieving for it, as the 15e queue is backed up a fair bit and C197s are on the big end of the 15e comfort zone. I'll do, say, A=30 on Q=25-100M because CADO is more efficient on low Q compared to ggnfs; that should be 15% of the job, give or take.

[VBCurtis]

500@16e8 finished. Starting poly select now.

Running 1e7-5e7 with P=6M and incr=4620. Leaving the low ad's for Sean & Ed to experiment with.

[EdH]

Not sure what value is wanted for this poly, but my Msieve suggests:

min E-value: 2.06e-15
expecting poly E from 7.98e-15 to > 9.18e-15

If I'm reading this correctly, this first post is shy of expectations, but well over the min (but, does Msieve E match cownoise E?):

n: 49166124173619026753031982944551526151417771554402090661062654848708821920824797208853190175969799268557269709859566937412206672909073923990203898320874417048790184714368584420136913755996169900623
skew: 125050816.141
c0: -100780658515871872306274606741432513729745330600
c1: -39547654921381624951406048498612939151914
c2: 429082982998007711345643654982747
c3: 7305269187829731646520191
c4: -14153532317442888
c5: -8837640
Y0: -102157892079785452666710108112736589239
Y1: 6898914490309534426789
# MurphyE (Bf=8.590e+09,Bg=4.295e+09,area=2.684e+16) = 7.256e-09
# f(x) = -8837640*x^5-14153532317442888*x^4+7305269187829731646520191*x^3+429082982998007711345643654982747*x^2-39547654921381624951406048498612939151914*x-100780658515871872306274606741432513729745330600
# g(x) = 6898914490309534426789*x-102157892079785452666710108112736589239

cownoise says:

215260513.93963    6.20323419e-15

[EdH]

I ran CADO-NFS with P=2M from 0-1e7 and P=12M for 0-1e6. I'm going to jump up to 2e8 and above for a while "just for fun."

[charybdis]

7000@1e9 completed. That should be enough ECM.

Might do some poly searching - what value of ropteffort do you guys recommend using?

[VBCurtis]

The root-opt phase is insignificant in time as a fraction of poly search time at this size, so I choose something big; I don't think there's any / much difference between 35 and 50, but I pick 40. 50 or 60 might search more on a small subset of polys, but not likely to improve things.

[charybdis]

Ran 5e7-15e7 with P=6M, same parameters as Curtis. This is the best I got:

n: 49166124173619026753031982944551526151417771554402090661062654848708821920824797208853190175969799268557269709859566937412206672909073923990203898320874417048790184714368584420136913755996169900623
skew: 10726805.758
c0: 62041855896374846734976480352559453307243919
c1: 86407605476371578679880507926689390979
c2: 1237169711766528739320735721169
c3: -2029521161120867427380351
c4: 625029182864520
c5: 1782742500
Y0: -62651154697255672749298406579454423016
Y1: 597457137027300338466379
# MurphyE (Bf=4.295e+09,Bg=4.295e+09,area=1.342e+16) = 5.807e-09

Cownoise score is 7.35059861e-15.