Reserve Exponent to Completion
 

Is there a way to reserve an exponent to completion? Meaning, if it has never been assigned to anyone, do the TF, P-1, ECM and LL just by putting in the exponent number?

An exponent is "complete" for the purposes of prime search either when a factor is found or when it has two matching LL residues. ECM is not normally part of this process as it is not cost-effective in CPU terms. Instead, we do ECM on small exponents which are already "complete" in an effort to find more factors.

If you take a DC assignment, it will by definition be "complete" when you have finished. If you take an LL assigment, you will automatically do any remaining TF and P-1 necessary on it, prior to starting the LL. It's not probably not a good idea for both the first-time LL and the DC to be done by the same user, and certainly not on the same machine.

If you take a TF or P-1 assignment and want to do the LL on it too, you could try changing the assigment in your worktodo file to "Test=...". This [I]might[/I] work, though I've never tried it. Bear in mind that you cant just change the "Test=" part. The parameters need to be changed as well.

thanks for the reply. Yea I was just thinking, for an exponent, if there's been no trial factoring on it started yet, I would want to start trial factoring before doing P-1 and eventually LL on it. Do I have to do all the steps manually? i.e. get manual assignment for trial factoring up to 63, then up to 64, then up to 65 etc and if none of these have a factor, then get a manual assignment for P-1 and if that doesn't have any factors, then get a manual assignment for LL or can I just say get a manual assignment for LL and it'll automatically go through the paces of trying the trial factoring first and then doing the P-1 and then will start LL? But from what you said here, it looks like if there has never been any TF and P-1 that it will automatically star the trial factoring on it and not jump straight to LL? " If you take an LL assigment, you will automatically do any remaining TF and P-1 necessary on it, prior to starting the LL."

[QUOTE=Unregistered;238798]Is there a way to reserve an exponent to completion? Meaning, if it has never been assigned to anyone, do the TF, P-1, ECM and LL just by putting in the exponent number?[/QUOTE]All exponents out to 999,999,999 have had some work done on them. First, all of the none prime exponents have been eliminated (this is trivial). And those above the very smallest have been already TF'ed to 64 bits or more. You can basically ask for a particular number on the Manual Assignments page, if you know how. If you ask for it as an LL and there is still TF and P-1 to do, you will do all of that too. There is another way, but that is another topic.

So, if you have a crazy formula that tells you what the next prime will be, you can ask for the exponent. If it is not already assigned, you should get it.

Thanks Uncwilly. I had another post that hasn't shown up yet, since I'm unregistered, but you answered my question. Mr. P-1 did as well, but I Just wasn't too sure until I read your reply.

Thanks

[QUOTE=Uncwilly;238842]...those above the very smallest have been already TF'ed to 64 bits or more.[/QUOTE]

Much of which was done in bulk, before the exponents were made available for assignment.

There's nothing to stop our guest from finding an unassigned exponent that has only been TF'ed to 64 bits, taking it as an LL assignment, and reducing the "TFed to" parameter, so that the client redoes the lover level TFs. This won't actually make a great deal of difference to the overall running time.

I'm not sure, however, that the client is even capable of TF at very low bit levels. I think there may be some, "this is definitely not necessary" code in there.

[QUOTE=Mr. P-1;238871]Much of which was done in bulk, before the exponents were made available for assignment.

There's nothing to stop our guest from finding an unassigned exponent that has only been TF'ed to 64 bits, taking it as an LL assignment, and reducing the "TFed to" parameter, so that the client redoes the lover level TFs. This won't actually make a great deal of difference to the overall running time.

I'm not sure, however, that the client is even capable of TF at very low bit levels. I think there may be some, "this is definitely not necessary" code in there.[/QUOTE]

No, it works as low as you could want. e.g. [URL="http://www.mersenne.org/report_exponent/?exp_lo=100043&exp_hi=&B1=Get+status"]M100043[/URL] has a factor at the lowest possible factor, with k=1:
q=2*1*100043+1=200087
And Prime95 finds it. With:
Factor=100043,0,40
It works:
[Work thread Nov 27 08:24] Starting trial factoring of M100043 to 2^40
[Work thread Nov 27 08:24] M100043 has a factor: 200087

[QUOTE=Mini-Geek;238887]No, it works as low as you could want. e.g. [URL="http://www.mersenne.org/report_exponent/?exp_lo=100043&exp_hi=&B1=Get+status"]M100043[/URL] has a factor at the lowest possible factor, with k=1:
q=2*1*100043+1=200087
And Prime95 finds it. With:
Factor=100043,0,40
It works:
[Work thread Nov 27 08:24] Starting trial factoring of M100043 to 2^40
[Work thread Nov 27 08:24] M100043 has a factor: 200087[/QUOTE]

It works NOW, but some of the previous versions of Prime95 just stopped at the first factor. As the program worked on different congruence classes, the first factor found was not necessarily the smaller.

Luigi