Factoring of Leftover Exponents

GP2 · 2003-10-13, 18:26

Well, for what it's worth, there are about 200 exponents in the range 10M - 20M that are factored to fewer bits than expected.

Filtering out the 15 exponents that are already currently assigned leaves 185 exponents between 13.4M - 17.6M that have been factored to 64 bits, but need to be factored to 65 bits.

In fact, most are between 13.4M - 15.6M, except for these:

Factor=16239491,64
Factor=16241977,64
Factor=16299971,64
Factor=16791539,64
Factor=17133169,64
Factor=17303413,64
Factor=17599927,64

Anybody with a P4 want these? Should I post the entire set of 185?

garo · 2003-10-13, 19:30

If those numbers have already been tested once or are in the process of being LL tested then there is no need to factor them to the last bit.

George had made a change to Primenet a while back that a number needs to be factored to one bit less if it has already been tested once as finding a factor would save one and not two LL tests.

GP2 · 2003-10-13, 20:13

Quote:

Originally posted by garo
If those numbers have already been tested once or are in the process of being LL tested then there is no need to factor them to the last bit.

OK, well that would disqualify all of those 185 exponents then.

garo · 2003-10-14, 00:02

One thing we could do, GP2, is to look for all exponents above 17.85M that have been LL tested but have only been factored to 64 bits or less. Those exponents can definitely be given to P4s.

The 17.85 limit is the point at which we go from 65 to 66 bits. I had checked a long while ago that there are no exponents under 17.85M that have not been factored to at least 1 bit less than the Prime95 limit. There were a few left which I did factor completely. Unfortunately I found no new factors

PS: Topic split off from original thread.

GP2 · 2003-10-14, 03:45

Quote:

Originally posted by garo
One thing we could do, GP2, is to look for all exponents above 17.85M that have been LL tested but have only been factored to 64 bits or less. Those exponents can definitely be given to P4s.

There's nothing below 20M that isn't currently in status.txt or cleared.txt.

I see about 550+ exponents in the 20M - 21M range however.

This is made trickier, though, by the fact that there is no "cleared" list for exponents that have had their factored-bits-depth increased without a factor being found.

The nofactor data file could be up to a week old. If it shows an exponent factored to N < 66 bits, it's possible that in the past few days someone factored it to 66 bits but didn't find a factor, and this will be reflected in the next weekly update of the nofactor file... but in the meantime the exponent has disappeared from status.txt (assignments) and there's nothing in cleared.txt to reflect it.

If you go beyond the leading edge of trial-factoring (25M+) there's a ton of exponents below 64 bits, tens of thousands in each million range. For anything between 20M and 25M you'd have to be careful and maybe only grab exponents very soon after the latest weekly release of the nofactor file.

garo · 2003-10-14, 06:18

Aha! I get what you are saying. I think that is why I decided to stop my leftover trial factoring at the point I did.

In any case, we should only release for trial factoring exponents that have had one LL test done and inadequate trial factoring done.

If there are no exponents that meet that criteria then we have less work!!

2003-10-13, 18:26	#1
GP2 Sep 2003 2×5×7×37 Posts	Well, for what it's worth, there are about 200 exponents in the range 10M - 20M that are factored to fewer bits than expected. Filtering out the 15 exponents that are already currently assigned leaves 185 exponents between 13.4M - 17.6M that have been factored to 64 bits, but need to be factored to 65 bits. In fact, most are between 13.4M - 15.6M, except for these: Factor=16239491,64 Factor=16241977,64 Factor=16299971,64 Factor=16791539,64 Factor=17133169,64 Factor=17303413,64 Factor=17599927,64 Anybody with a P4 want these? Should I post the entire set of 185? Last fiddled with by GP2 on 2003-10-13 at 18:27

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Could we have the ability to re-reserve exponents for P-1 factoring?	ixfd64	PrimeNet	21	2021-01-24 18:13
How many DC exponents are GPU Factoring planning to reserve?	patrik	GPU to 72	8	2013-03-13 16:03
factoring while dealing with exponents	secretdude	Homework Help	4	2007-12-13 04:56
Computer Factoring 2 Exponents Simultaneously?	jinydu	Lounge	13	2005-03-07 01:37
ERROR 7:Server has run out of exponents to assign(factoring)	dsouza123	Software	4	2003-01-14 11:10

2003-10-13, 19:30	#2
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	If those numbers have already been tested once or are in the process of being LL tested then there is no need to factor them to the last bit. George had made a change to Primenet a while back that a number needs to be factored to one bit less if it has already been tested once as finding a factor would save one and not two LL tests.

2003-10-14, 00:02	#4
garo Aug 2002 Termonfeckin, IE 2768₁₀ Posts	One thing we could do, GP2, is to look for all exponents above 17.85M that have been LL tested but have only been factored to 64 bits or less. Those exponents can definitely be given to P4s. The 17.85 limit is the point at which we go from 65 to 66 bits. I had checked a long while ago that there are no exponents under 17.85M that have not been factored to at least 1 bit less than the Prime95 limit. There were a few left which I did factor completely. Unfortunately I found no new factors PS: Topic split off from original thread.

2003-10-14, 06:18	#6
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	Aha! I get what you are saying. I think that is why I decided to stop my leftover trial factoring at the point I did. In any case, we should only release for trial factoring exponents that have had one LL test done and inadequate trial factoring done. If there are no exponents that meet that criteria then we have less work!!