Polynomial Request Thread

[swellman]

Originally Posted by VBCurtis

That's a pretty good poly score for a composite this size. I think you can just run with that one!

Thanks! It was a borderline SNFS/GNFS job, guess we will go with SNFS. Appreciate the advice.

[swellman]

Can someone with a GPU rig please help me find a poly for this job?

Originally Posted by XYYXF

С142_131_124 needs GNFS:

5210529691066565797257456228658139553918217336357339442709049302999626656196741161542669359479005591654809933254107274624028909150290695297551

I'm happy to run the GNFS job but with my CPUs such poly searches take forever. Thank you in advance.

[wombatman]

I got you. Gonna use the 750Ti on my work computer, though, so I have absolutely ZERO idea how long it might take. Regardless, I'll post something up in the next couple of days.

[swellman]

Thanks!

[wombatman]

Turns out a C142 flies on the 750Ti, so I've actually gone through the entire recommended range. Currently in the root finding stage, but I don't have MinGW (and thus no sorting available), so I'm just running -npr on the entire ms file. Probably will have something by tomorrow?

[Dubslow]

Originally Posted by wombatman

so I'm just running -npr on the entire ms file. Probably will have something by tomorrow?

That's a terrible idea, and will take easily x100 times as long to do. -npr is extremely intensive compared to the first two stages. You really need a narrow it down to the x hundred best candidates. As in, better to pastebin and link here and have someone sort it for you.

[wombatman]

Oh, I'm well aware that it's not the proper thing to do--this was purely laziness. If I come in tomorrow and it's still going, I'll just send it to myself and sort/re-run with the top 100-200 hits.

[Dubslow]

Originally Posted by wombatman

If I come in tomorrow and it's still going,

We're talking at least as long as the sieving itself if not longer.

[VBCurtis]

Originally Posted by wombatman

Turns out a C142 flies on the 750Ti, so I've actually gone through the entire recommended range. Currently in the root finding stage, but I don't have MinGW (and thus no sorting available), so I'm just running -npr on the entire ms file. Probably will have something by tomorrow?

You can get away without sorting by setting stage2norm to be about 1/30th the default size. This will produce 100-200 hits/day on a 750ti, on which rootsieve takes an hour or two.

[wombatman]

Originally Posted by Dubslow

We're talking at least as long as the sieving itself if not longer.

polynomial selection complete
R0: -1218423089286444403180113917
R1: 23258540774071
A0: 3594145623147156846353534367557400
A1: 73328877321934102656644258040
A2: -308579838389273978193608
A3: -853957752589087195
A4: 1705125392418
A5: 1940400
skew 423678.78, size 9.496e-014, alpha -7.391, combined = 1.603e-011 rroots = 5
elapsed time 02:10:35

(Kidding around, of course. I had dropped the stage 1 and 2 norms by 1/10 of default choice and limited the minimum e-score to 1.4e-11, so it was probably able to complete more quickly than usual.)

[Dubslow]

Originally Posted by wombatman

(Kidding around, of course. I had dropped the stage 1 and 2 norms by 1/10 of default choice and limited the minimum e-score to 1.4e-11, so it was probably able to complete more quickly than usual.)

Critical information that