[QUOTE=henryzz;478870]If factors/day is your aim then you should probably run curves at a higher level. It will take longer to complete 25 digits but you will find a lot more factors >25 digits and there will be less to do for 30 digits.[/QUOTE]

This. Try some intermediate values of B1, e.g. run a day or two at B1 = 100k and see how your rate of factors-found changes. I expect you'll find nearly as many factors, while more of them will be 28-to-32 digits than before. If so, go to 150k for a couple days, repeat until the rate drops off or you get bored.

A SNFS 200 quartic is bad because the rational norm is much larger than the algebraic norm. One thing I've wondered is if it's possible to use a quadratic (degree 2) on the rational side instead of the usual linear (degree 1).

I've heard of using two quadratics for GNFS, one on each side, but I can't remember where. It wasn't competitive with the normal way though.

Chris

Here are two more full factorizations from the t2100 file.

C532 = P32 * PRP501 [url]http://factordb.com/index.php?id=1100000000438624599[/url]

C653 = P33 * PRP621 [url]http://factordb.com/index.php?id=1100000000450836049[/url]

t2100
 

31^257-1 is divisible by p50.
p50 = 35890392908870171521864015727558211438952448644537

Here are two more full factorizations from the t2100 file:

C659 = P30 * PRP630 [url]http://factordb.com/index.php?id=1100000000655014354[/url]

C862 = P27 * PRP836 [url]http://factordb.com/index.php?id=1100000000685532706[/url]

I just noticed that there are now less than 65,000 composites in the t2100 file. Great work everyone! :banana:



Also, here's another full factorization from the t2100 file:

C826 = P27 * PRP800 [url]http://factordb.com/index.php?id=1100000000615521636[/url]

I have a bunch in a long log file if anyone wants to sort through them. My 1 curve at 1e6 run found a lot

Trying my hand at some GNFS:

[URL="http://www.factordb.com/index.php?query=459913866740209041331349460680262230936475470929114765151226584224482455774571483444467129409276961%5E3-1"]459913866740209041331349460680262230936475470929114765151226584224482455774571483444467129409276961^3-1[/URL]

Seems like composites around 140 - 150 digits should take a couple of days to a week for me.

[QUOTE=henryzz;479654]I have a bunch in a long log file if anyone wants to sort through them. My 1 curve at 1e6 run found a lot[/QUOTE]

If you post it here I'll go through it

[QUOTE=Wick;479659]If you post it here I'll go through it[/QUOTE]

[url]https://www.sendspace.com/file/gd8bjo[/url]
I went through upto 555 digits above.

[QUOTE=henryzz;479662][url]https://www.sendspace.com/file/gd8bjo[/url]
I went through upto 555 digits above.[/QUOTE]

all reported 52 new factors ow which 2 FF