[QUOTE=pinhodecarlos;282485]What am I doing wrong? Can someone explain me step by step how to run polynomial search?[/QUOTE]

The parameters for very large inputs have (at present) to be set by changing the file gnfs/poly_skew.c in the source code and recompiling (make x86_64 on linux, on Windows I have no idea); if you're not happy with recompiling msieve then I'm afraid there's not much opportunity to help out with this particular project at this very early stage.

If anyone's interested in tweaking the stage-2 parameters for M1009

840 1532426581029323681171 14353279785776501210693417737211891227753

is the best of ten thousand stage-1 hits I've found so far (stage-2 root score 4e+27 as against a median of 1.7e+29) and the source of the 1.919e-15 score. Stage-2 root score 2.4e+28 is the 99th-percentile.

(I don't know whether the new stage 2 has changed the double role of max_norm; for the 197-digit polynomial search I found it seemed to help to compute the norm for every polynomial, cut down to the top 1%, then run -np2 on that top 1% with max_norm set much larger. But I'm not sure that wasn't credibly dismissed as voodoo even at the time)

The new stage 2 works exactly the same as before, it's just much more robust when choosing degree-6 polynomials.

[QUOTE=fivemack;282496]The parameters for very large inputs have (at present) to be set by changing the file gnfs/poly_skew.c in the source code and recompiling (make x86_64 on linux, on Windows I have no idea); if you're not happy with recompiling msieve then I'm afraid there's not much opportunity to help out with this particular project at this very early stage.[/QUOTE]

I don't know how to compile but I would really like to help out (windows 32 and 64 bits version).

[QUOTE=fivemack;282496]The parameters for very large inputs have (at present) to be set by changing the file gnfs/poly_skew.c in the source code and recompiling (make x86_64 on linux, on Windows I have no idea); if you're not happy with recompiling msieve then I'm afraid there's not much opportunity to help out with this particular project at this very early stage.[/QUOTE]

Could you give me an exact description of the files that need to be changed and the exact details of the changes needed?

Or just wait until msieve 1.50 is out, which has params for this composite and you won't have to compile anything.

I'm running -np1 1000,5000.

I've done (very nearly) 20k@260M on 3,589- c213. This one should be ready for SNFS if you are interested.

That's a lot of curves, but I'm afraid I have an irrational preference for Mersenne numbers; I've reserved M929 for the forum with Wagstaff and am trying to get good parameters (in particular, to see if I can get it to be a 15e job)

[QUOTE=fivemack;283053]That's a lot of curves, but I'm afraid I have an irrational preference for Mersenne numbers; I've reserved M929 for the forum with Wagstaff and am trying to get good parameters (in particular, to see if I can get it to be a 15e job)[/QUOTE]

Why would the Mersenne forum have a preference for such numbers?
I can't imagine.

[QUOTE=R.D. Silverman;283061]Why would the Mersenne forum have a preference for such numbers?
I can't imagine.[/QUOTE]Ooh! Irony, and from an American too.

That's something you don't see every day.