skipped bit levels

[tha]

I redid P-1 on this exponent with higher B1 and B2 bounds. A factor of just over 65 bits was found, and according to the server data this bit level was skipped for some reason. Can we compile a worktodo file with all skipped bit levels for all exponents?

[Madpoo]

Originally Posted by tha

I redid P-1 on this exponent with higher B1 and B2 bounds. A factor of just over 65 bits was found, and according to the server data this bit level was skipped for some reason. Can we compile a worktodo file with all skipped bit levels for all exponents?

It's problematic because the data isn't clear-cut on which bit levels are *actually* skipped versus which ones only appear to be skipped because the client didn't include a starting bit-level.

We've looked into this previously in another thread and I've checked the data... basically there are a lot of entries where it appears that bits may have been skipped if we relied on the data alone, but knowing that some early code didn't include "from 2^x" in the results means that we have no idea where it started, only that it ended up at "to 2^x" bit level.

Trying to double-check the TF on all of those where it was probably done previously would be like looking for needles in haystacks where you're not really sure there's a needle in there at all.

[Mark Rose]

Originally Posted by Madpoo

It's problematic because the data isn't clear-cut on which bit levels are *actually* skipped versus which ones only appear to be skipped because the client didn't include a starting bit-level.

We've looked into this previously in another thread and I've checked the data... basically there are a lot of entries where it appears that bits may have been skipped if we relied on the data alone, but knowing that some early code didn't include "from 2^x" in the results means that we have no idea where it started, only that it ended up at "to 2^x" bit level.

Trying to double-check the TF on all of those where it was probably done previously would be like looking for needles in haystacks where you're not really sure there's a needle in there at all.

Yeah, but in the case above, one bit level was skipped according to the records.

[Mark Rose]

I found another exponent with a missing bit level: M76903093.

So I did the missing work, and submitted the results:

no factor for M76903093 from 2^62 to 2^64 [mfaktc 0.21 75bit_mul32_gs]

and got:

processing: TF no-factor for M76903093 (2⁶²-2⁶⁴)
Error code: 40, error text: TF result for M76903093 was not needed

Which is not true, because there is a gap.

[tha]

Originally Posted by Mark Rose

processing: TF no-factor for M76903093 (2⁶²-2⁶⁴)
Error code: 40, error text: TF result for M76903093 was not needed

Which is not true, because there is a gap.

According to the server data you also did that exponent today from 2^73 to 2^74. Is that correct?

[Mark Rose]

Originally Posted by tha

According to the server data you also did that exponent today from 2^73 to 2^74. Is that correct?

Yes.

[ATH]

Originally Posted by Madpoo

It's problematic because the data isn't clear-cut on which bit levels are *actually* skipped versus which ones only appear to be skipped because the client didn't include a starting bit-level.

We've looked into this previously in another thread and I've checked the data... basically there are a lot of entries where it appears that bits may have been skipped if we relied on the data alone, but knowing that some early code didn't include "from 2^x" in the results means that we have no idea where it started, only that it ended up at "to 2^x" bit level.

Trying to double-check the TF on all of those where it was probably done previously would be like looking for needles in haystacks where you're not really sure there's a needle in there at all.

What if you only look at exponents above 34M that needs a DC or LL?

[Madpoo]

Originally Posted by Mark Rose

...one bit level was skipped according to the records.

Yeah, and there's the problem... the records for TF may be erroneous, as in it may not show that someone actually did the TF in a certain range.

I don't know why some TF users may have skipped factoring in a certain range. Like in the example you provided at the start, there's an entry for:
"no factor to 2^65"
and then one for:
"no factor from 2^66 to 2^67"

Did Nesowa really skip 2^65 to 2^66 or did they run it and it didn't get reported in? I mean, logically, yeah, the entry *should* be there saying it was done in that range, but some clients didn't seem to do a great job of reporting stuff, especially if they were manually sending results and just grabbed the last thing it did... who knows.

Plus as you've seen, going back and trying to fill in the gaps (if such gaps actually do exist in reality, not just from the reporting perspective) has problems since the server won't accept TF work being done at a lower bit level than whatever is currently the highest bit level that someone has checked in.

Someone may have gone through and "factored the gaps" for some exponents but there's no record since the server isn't accepting them.

Whether Primenet should be modified to accept these "less than max" TF results, I don't know. I guess it couldn't hurt, but someone could take advantage of that to do a bunch of small TF work and hope for a lot of credit by doing tons of work over again, like factoring everything from 2^2 through 2^50 and flooding with results. Maybe it would make sense to accept it but not give any credit besides a nominal amount? Beats me... I'm just thinking of ways someone would take advantage of that kind of thing.

[Mark Rose]

Even if no credit were given, it would be good to track that the work is done.

Could the line be added to the history at least?

[retina]

Originally Posted by Madpoo

Whether Primenet should be modified to accept these "less than max" TF results, I don't know. I guess it couldn't hurt, but someone could take advantage of that to do a bunch of small TF work and hope for a lot of credit by doing tons of work over again, like factoring everything from 2^2 through 2^50 and flooding with results. Maybe it would make sense to accept it but not give any credit besides a nominal amount? Beats me... I'm just thinking of ways someone would take advantage of that kind of thing.

Credit costs nothing. It is just a number. Let people take credit if they wish to. It doesn't matter. Besides, factoring the "gaps" from 1 to 2^64 would be very quick, it is a one time event, and giving credit by the normal computation wouldn't amount to much anyway.

[Mark Rose]

I may crawl the exponents and see just how much work there is. I've been refreshing my Java by playing around with PrimeNet XML. This could be my next project.