yafu.pl Usage Currently

[EdH]

Originally Posted by yoyo

You should factor numbers beginning with 70 digits. This takes longer and the probability to hit the request limit per hour is much lower.

That is/was my intention. But, I thought I'd turn some machines toward the massive addition of all the 60 digit composites. Maybe I moved too many machines...

[LaurV]

You should actually concentrate to "over 80", or even "over 90", right now. Everything under that is "insignificant" in quantity, and under 70 is anyhow handled by the local workers. So the factoring time would take longer and you should not hit into that limit.

[EdH]

Originally Posted by LaurV

You should actually concentrate to "over 80", or even "over 90", right now. Everything under that is "insignificant" in quantity, and under 70 is anyhow handled by the local workers. So the factoring time would take longer and you should not hit into that limit.

I'm actually running a lot of 98s, now that I'm using 1000 for my random value.

On another note, does YAFU multithread anything besides SIQS and NFS?

If the above answer is, "No," would it be more efficient to run four separate instances of YAFU single-threaded, than one instance with four threads for these composites?

[EdH]

Originally Posted by EdH

...
On another note, does YAFU multithread anything besides SIQS and NFS?

If the above answer is, "No," would it be more efficient to run four separate instances of YAFU single-threaded, than one instance with four threads for these composites?

I found out, via top, that the answer is, "Yes." ECM is also multi-threaded.

Another subject: The db is almost .5M further behind than when I started running yafu.pl. Are there times when the number of unfactored composites actually decreases, or does that require some heavy machinery to come in for a bit?

[EdH]

I have changed over to Antonio's script and it is running well. I look forward to seeing how the on-the-fly variables adjust, but having had a power hit that knocked out most of my machines, at least one fatally, I wonder about creating persistent values...

I'm also ever so disappointed that Antonio's script doesn't provide all the feedback from YAFU, either. I (somewhat) understand the trouble of YAFU not sending CRs so it can overwrite the lines, but there should be a way to work that into the code. Maybe I'll modify my YAFU code...

Anyway, thanks Antonio. Your script is working fine straight from the d/l.

I did notice the cookie variable, but I can't locate my factordb cookie on any of my machines and don't know enough about them to really use that feature, yet. More things to study...