P-1 factorer

[smh]

As I see it, there are two immediate areas that should be factored. n=510,000 and n=4,000,000.

Numbers around N=500K already had a lot of sieving done in comparision to the time it takes to prp them. If you spend the optimal time sieving these low numbers you only have the time to sieve with low bounds, which makes the chances of finding a factor above the current sieve depth low.

For numbers around N=4M this is completely different. The optimal time to prp should be something like the time it takes to prp a number twice (and add a little bit of time for faulty tests which need a tripple check) divided by the chance to find a factor with P-1 using the optimal bounds. (this doesn't take into account the sieving time on each number.

This makes me wonder, does anyone have an idea of how many giga hertz days/weeks have been put into sieving? I wonder how much time this is per exponent.

3) Test the new SBFactor. It works now and is fast. All the low hanging P-1 fruit is still in the orchard for those who want to pick it. (smh/biwema/rosebud, I'm looking at you )

At least it finds the two factors it missed with the old version.

I'll plan more p-1 when there is a lists with numbers available.

1590402916048273 = 2 ^ 4 x 3 x 23 x 43 x 197 x 1381 x 123143 + 1

I found this with B1=50000 B2=5000000 (both a little overkill).

This is about as smooth of factor as you get. 2^4!? That's beging to be uncovered with P-1.

2^4 is only 16, so it hardly makes any difference. A few above 1000 and a large one in the second stage is what gives you the large factors.