Poachers - Page 2

tha · 2014-03-17, 23:17

Quote:

Originally Posted by lycorn

Just out of curiosity, what´s your motivation for those assignments in particular?

The general idea among a substantial amount of participants is that a factor is worth more than two matching LL tests. The range 0 to 10M is substantially better trial factored than the higher ranges. In the many graphs you can find a significant dip from 10M to a few million higher. I noticed some GPU owners are doing trial factoring in that range. So P-1 in that area would contribute.

So I started at this page of the mersenne site of James and work from there on. Work is not coordinated, so use the LMH subforum to claim your non server assigned work.

lycorn · 2014-03-19, 17:26

Yeah, got your idea. Actually, I´ve also been bit by the "finding factors for small DCed exponents" bug. I´ll probably carry on for some more time performing ECM/TF/P-1 on small exponents.
But I didn´t understand what you said about 0-10M range being "better trial factored" than subsequent ranges. In fact, the average bit level for that range is way lower than 10-20M (which is no surprise, given the significantly higher difficulty of digging deeper in the 0-10M than in the 10-20M).
Or are you already takimg this into account when you say it´s "better TFed"?

2014-03-19, 17:26	#13
lycorn "GIMFS" Sep 2002 Oeiras, Portugal 2702₈ Posts	Yeah, got your idea. Actually, I´ve also been bit by the "finding factors for small DCed exponents" bug. I´ll probably carry on for some more time performing ECM/TF/P-1 on small exponents. But I didn´t understand what you said about 0-10M range being "better trial factored" than subsequent ranges. In fact, the average bit level for that range is way lower than 10-20M (which is no surprise, given the significantly higher difficulty of digging deeper in the 0-10M than in the 10-20M). Or are you already takimg this into account when you say it´s "better TFed"?