TF Factor Found CPU Credit

TheMawn · 2013-06-16, 19:17

I've been getting about 30GHz-Days per exponent on mfaktc, doing 60M from 73 to 74. I found a couple of factors recently, and the CPU credit I got was roughly 60GHz-Days.

Why do I get more credit when I find a factor than when I don't? Surely the calculations take less time if they are stopped early?

firejuggler · 2013-06-16, 21:12

another bit TF'ed doesn't really give any info, it still can be prime or composite.
A factor, on the other side eliminate an exponent for sure.

TheMawn · 2013-06-17, 02:34

But 60GHz-Days isn't enough credit to cover the LL-test I saved. And the double check.

NBtarheel_33 · 2013-06-17, 06:21

It has nothing to do with GHz-days saved; rather, this is a known bug when manually submitting TF factor-found results. It is relatively simple to fix (say a factor is 73.63 bits, then just credit .63 of the full credit from 73 to 74) but simply has not been a priority item for the PrimeNet administrators.

You'll notice that (manually submitted) TF factors are credited as though they were found by P-1, and some (manually submitted) P-1 factors are credited as though they were found by ECM. The manual submission process will also sometimes accept a P-1 that was done to bounds B2 > B1, but record (and credit) it as though it was only done to B1.

All of this is very easy to fix, but it's the kind of thing that ends up on the back burner in favor of other more interesting/important tasks, e.g. code optimization.

2013-06-16, 19:17	#1
TheMawn May 2013 East. Always East. 11·157 Posts	TF Factor Found CPU Credit I've been getting about 30GHz-Days per exponent on mfaktc, doing 60M from 73 to 74. I found a couple of factors recently, and the CPU credit I got was roughly 60GHz-Days. Why do I get more credit when I find a factor than when I don't? Surely the calculations take less time if they are stopped early?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Factor found that should have been found by P-1	tha	Data	65	2020-08-05 21:11
F12 factor found?	johnadam74	FermatSearch	16	2016-11-03 12:10
mfaktc TF credit 2x higher if factor found?	S34960zz	PrimeNet	10	2011-10-13 07:00
TF factor credit	TimSorbet	PrimeNet	2	2009-08-28 18:37
Who gets credit for first 10M prime if 2 are found,same time	eepiccolo	Lounge	20	2003-08-03 02:23

2013-06-16, 21:12	#2
firejuggler "Vincent" Apr 2010 Over the rainbow 2²×7×103 Posts	another bit TF'ed doesn't really give any info, it still can be prime or composite. A factor, on the other side eliminate an exponent for sure.

2013-06-17, 02:34	#3
TheMawn May 2013 East. Always East. 11·157 Posts	But 60GHz-Days isn't enough credit to cover the LL-test I saved. And the double check.

2013-06-17, 06:21	#4
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 2133₈ Posts	It has nothing to do with GHz-days saved; rather, this is a known bug when manually submitting TF factor-found results. It is relatively simple to fix (say a factor is 73.63 bits, then just credit .63 of the full credit from 73 to 74) but simply has not been a priority item for the PrimeNet administrators. You'll notice that (manually submitted) TF factors are credited as though they were found by P-1, and some (manually submitted) P-1 factors are credited as though they were found by ECM. The manual submission process will also sometimes accept a P-1 that was done to bounds B2 > B1, but record (and credit) it as though it was only done to B1. All of this is very easy to fix, but it's the kind of thing that ends up on the back burner in favor of other more interesting/important tasks, e.g. code optimization.