20211105, 02:47  #980 
Apr 2020
1041_{8} Posts 
Depends how much detail you want
NFS has three stages. 1. Polynomial selection. You need to find two polynomials f(x) and g(x) that have a common root modulo N, the number you want to factor. In general smaller coefficients are better. Almost always g(x) (the "rational polynomial") will be linear, and f(x) (the "algebraic polynomial") will have higher degree, usually 4, 5 or 6. SNFS is when the form of the number naturally gives you such polynomials: for the number we're discussing here, these are 31479823396757*x^61 and x31479823396757^3, which have common root 31479823396757^3 mod N. GNFS is for numbers that don't have a special form, and the polynomials must be found by computer search. There are algorithms for this that are used by msieve and CADO. The "c6" and "c0" lines in the polynomial file are the coefficients of the algebraic polynomial; I omitted c5 to c1 because those are all zero. The rational polynomial coefficients are Y1 and Y0. 2. Sieving. This consists of finding some large number of "relations". In some sense these are like equations in a large number of variables, such that if enough are found, the resulting system of equations can be solved. I like LaurV's explanation here. There are lots of different parameters that control sieving: this is what rlim/alim, lpbr/lpba etc are doing. 3. Postprocessing. This is where the relations are combined together and the system of equations is solved. Each solution to the equations has a 50% chance of finding the factors. SNFS is much faster than GNFS for numbers of the same size (where, as I mentioned above, it's the size of the original number that matters for SNFS). A rough approximation is that SNFS of d digits is equivalent to GNFS of 0.69*d digits, but this doesn't hold when the SNFS polynomial has large coefficients or the "wrong" degree for the size of the number. GNFS difficulty roughly doubles every 5.5 digits. This thread has some more detailed mathematical explanations if you're interested. These are very much not ELI5, though maybe some are ELI18. Last fiddled with by charybdis on 20211105 at 02:48 
20211105, 02:58  #981 
May 2003
106_{16} Posts 
Thanks. ELI18 may have to do.

20211109, 15:29  #982 
May 2003
406_{8} Posts 
I decided to start working through the t2200 file for now.
I finished (4704006414025977517737010367874433^7 1)/(47040064140259775177370103678744331) yesterday Wrapping up (164840458433247831545380402768066382125607140850350063^51)/(1648404584332478315453804027680663821256071408503500631) in about an hour If no one is working on it, (55861871484788877230142332670582003900341141786568666929551231846127870452424196830136166348296553690709766995700853633762202956955812747696231496996184865884037^11)/(558618714847888772301423326705820039003411417865686669295512318461278704524241968301361663482965536907097669957008536337622029569558127476962314969961848658840371) is next 
20211117, 03:14  #983 
May 2003
2·131 Posts 
Anybody working on (14275963305110219876269575868443201054709206526952694749229865347248013990032567493253^31)/(142759633051102198762695758684432010547092065269526947492298653472480139900325674932531)?

20211117, 09:35  #984 
Dec 2017
71 Posts 
Working on 865901231558393743256873067633607^71 from the t2200 file.
Does anyone have an eta on the mwrb2200 file? Last fiddled with by Brownfox on 20211117 at 09:37 Reason: Copy paste error 
20211118, 01:25  #985 
May 2003
2×131 Posts 
Any idea how we could get composites from the mwrb2100 file into the NFS@home project? Looks like they are mostly doing Cunningham numbers.

20211118, 05:17  #986 
"Curtis"
Feb 2005
Riverside, CA
11713_{8} Posts 
Have you looked at the nfs@home subforum?
ECMpretest a candidate to at least the usual depth, testsieve on the appropriate siever to determine Qrange, and post the job. The size and parameter restrictions for each queue are posted at the top of each job thread. 14e jobs are discouraged, as those can be done by individuals. We could use a few jobs on 15esmall and 15e queues, as the supply of fully ECM'ed jobs has been thin this fall. The 16e queue has plenty of supply, so jobs bigger than SNFS280/GNFS190 should maybe not be a priority just now. 
20211120, 04:05  #987 
May 2003
2·131 Posts 
Running ECM on 4125099241766702497828754545051526255358191877622953835963928559535525025242323612793829455850360980990732729168762413572946726127^31
from the t2200 file Last fiddled with by ThomRuley on 20211120 at 04:05 
20211125, 00:54  #988 
May 2003
106_{16} Posts 
NFS on 360980990732729168762413572946726127^31

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Passive Pascal  Xyzzy  GPU Computing  1  20170517 20:22 
Tesla P100 — 5.4 DP TeraFLOPS — Pascal  Mark Rose  GPU Computing  52  20160702 12:11 
Nvidia Pascal, a third of DP  firejuggler  GPU Computing  12  20160223 06:55 
Calculating perfect numbers in Pascal  Elhueno  Homework Help  5  20080612 16:37 
Factorization attempt to a c163  a new Odd Perfect Number roadblock  jchein1  Factoring  30  20050530 14:43 