20100615, 09:09  #1 
"Nathan"
Jul 2008
Maryland, USA
5·223 Posts 
Successful TF worth more than unsuccessful TF?!
I'm working on bringing the whole 930931M range to 2^64, reporting results manually. Just now, I uploaded a bunch of results, and observed that M930137627 (which had a factor, namely 9440785022214022409) has been credited 0.0032 GHzdays, despite the fact that thousands of assignments in this range that have been unsuccessful have all been credited 0.0021 GHzdays. Obviously, if we find a factor, we do less work and spend less CPU time than if we have to go all the way to 2^64. So how can a successful TF be worth 1.5 times an unsuccessful TF on the same range?
Note that every other time I've found factors, the CPU credit (as expected) has been less than 0.0021 GHzdays (indeed, one quick kill actually was below the 4decimal threshold: reported credit was 0.0000). One other note that might be important  I accidentally sent the latest batch of results (including this factor) without signing in, so everything went to ANONYMOUS. But since I had assignment keys, I went ahead and signed in and resubmitted everything  the system kicked the results out as "not needed" but I got the proper credit. I can't imagine why this would cause the higherthanmaxcredit anomaly, though. BTW this range has a decent vein of factors  from M930000000 to M930137627, I've found 65 factors out of 3,327 tries. That's 1 in 51.185, when we expect 1 in 64. 
20100615, 12:07  #2 
Account Deleted
"Tim Sorbera"
Aug 2006
San Antonio, TX USA
11×389 Posts 
According to http://mersennearies.sili.net/credi...tton=Calculate, the credit for that factor should be 0.001456 GHzdays, (which is more like what you'd expect) and the credit for an unsuccessful TF of 2^63 to 2^64 is 0.002111 (matching the 0.0021 that you see).
I don't know why it would give 0.0032 GHzdays instead. If it was exactly double the calculated value, I'd guess doublecrediting due to the unusual circumstances, but how it is I really don't know. Last fiddled with by TimSorbet on 20100615 at 12:08 
20100615, 15:17  #3 
Aug 2002
Termonfeckin, IE
2768_{10} Posts 
I think there was a 2.3 times bonus for factors found in Primenet v4. I thought it had been removed for v5.

20100615, 15:37  #4 
Account Deleted
"Tim Sorbera"
Aug 2006
San Antonio, TX USA
11×389 Posts 
I don't think such a bonus exists now, since otherwise factors having more credit than their ranges would be quite commonplace. Besides, 0.001456*2.3=0.0033488 which would round to 0.0033, not 0.0032.
Last fiddled with by TimSorbet on 20100615 at 15:37 
20100615, 22:55  #5 
"Nathan"
Jul 2008
Maryland, USA
5×223 Posts 
There can't be a bonus...
...or else it seems like I would have other factors with credit above 0.0021, and there have been none. One came close at 0.0020, but nothing above 0.0021 except this lone guy. I wonder if something may have gone amiss since the factor got reported once by ANONYMOUS, and then once by myself with the proper assignment key. Maybe PrimeNet thinks that the factor is in the 6465 range instead of 6364? That would account for what is basically a doubling in the expected credit.

20100617, 00:17  #6 
"Nathan"
Jul 2008
Maryland, USA
5·223 Posts 
M930163667 gives max credit, as expected
M930163667 has the factor 15205068979981957919 = 119 (mod 120), which is the last residue class that Prime95 checks in the TF process. Hence you'd expect to receive nearly the same credit as an unsuccessful TF, which I did  namely 0.0021.
Note that the problem child, M930137627, had the factor 9440785022214022409 = 89 (mod 120), which is towards, but not at the end of the TF process. So this isn't even at the max credit level, let alone above it. Something weird definitely happened there. 
Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Is it worth persisting with this machine?  stebbo  Hardware  6  20170527 04:05 
Is it worth keeping the relations?  fivemack  Factoring  2  20130519 18:40 
just finished the biggest (successful) LL test  joblack  Lounge  35  20110213 02:00 
Does it worth it?  victor  Lounge  30  20090530 21:53 
What makes a team successful?  eepiccolo  Teams  5  20030524 23:50 