BOINC effort for CRUS

[gd_barnes]

Originally Posted by VBCurtis

R19 seems like an interesting base for BOINC work in the medium term. Once it is complete to 100k, would BOINCers test 100k-250k if I make a sieve file? It would take me 30ish days to sieve fully, I expect. Is that worthwhile?

Thanks for the offer. I'm not stopping the BOINC effort from doing what it wants but I'd prefer that it not take on such a large effort for a base with no chance of proof in our lifetimes. After some of the smallish bases < 50 with large #'s of k's remaining have been searched to n=100K, I feel that the huge resources needed to search them higher would be better utilized in some of our high-n searches on bases with few k's remaining that have a reasonable chance of being proven. If you'd like to help sieve, consider taking our Sierp base 6 n=1M-2M file (12 k's). A link is on our reservations page. It is already sieved to P=220T. That would be the kind of effort I'd like to see BOINC go after.

[Siemelink]

Originally Posted by MyDogBuster

BTW: A nice job being done by all on BOINC. Over 200 active users and 4200 active ports. Zowza!!!!!

Hear hear!

[VBCurtis]

Gary-
I prefer to help on Riesel bases. Does R30 fit your criteria? 9 k's, at 500k now, unreserved. If I built a sieve file for 500k-1M, would BOINC test it?

[rebirther]

Originally Posted by VBCurtis

Gary-
I prefer to help on Riesel bases. Does R30 fit your criteria? 9 k's, at 500k now, unreserved. If I built a sieve file for 500k-1M, would BOINC test it?

How long is the runtime?

[gd_barnes]

Originally Posted by VBCurtis

Gary-
I prefer to help on Riesel bases. Does R30 fit your criteria? 9 k's, at 500k now, unreserved. If I built a sieve file for 500k-1M, would BOINC test it?

Close but how about R28 instead? 4 k's for n=500k-1M. It's currently tested to n=450k by Ian and he should be done to n=500k anytime now. Another good one would be R37. 16 k's for n=100k-250k. I tend to look at the bases with lower relative difficulty (to catch them up so to speak) as shown at http://www.noprimeleftbehind.net/cru...s-unproven.htm.

Of course not stopping anyone from doing whatever he wants to do.

[MyDogBuster]

It's currently tested to n=450k by Ian and he should be done to n=500k anytime now.

Should be done in about 2 weeks.

[VBCurtis]

Originally Posted by gd_barnes

Close but how about R28 instead? 4 k's for n=500k-1M. It's currently tested to n=450k by Ian and he should be done to n=500k anytime now. Another good one would be R37. 16 k's for n=100k-250k. I tend to look at the bases with lower relative difficulty (to catch them up so to speak) as shown at http://www.noprimeleftbehind.net/cru...s-unproven.htm.

Of course not stopping anyone from doing whatever he wants to do.

OK. R37 is a smaller sieve project anyway, in case I lose interest before a 500k-1M sieve is done. I'll start this in a few days, 100-250k.

[rebirther]

For BOINC ranges between 25-200k are good at the moment. If someone want to sieve some bases in the list 10-25k or 100-200k with some smaller k this would be fine. So we can combine the power by sieving, starting new bases up to 10-25k and BOINC is testing 25-200k. At the moment the rate of 4.6k tests in progress are low. My goal is 10k. Iam always open for some wishes what bases are run first.

[VBCurtis]

Originally Posted by rebirther

For BOINC ranges between 25-200k are good at the moment. If someone want to sieve some bases in the list 10-25k or 100-200k with some smaller k this would be fine. So we can combine the power by sieving, starting new bases up to 10-25k and BOINC is testing 25-200k. At the moment the rate of 4.6k tests in progress are low. My goal is 10k. Iam always open for some wishes what bases are run first.

Are you suggesting I sieve 100-200k rather than 100-250? In the long run, this is fine, as the next sieve can be 200-500k to hit the usual next milestone. Note that base 37 is pretty low, so the tests will run quite a bit faster than bases over 100- meaning 250k isn't unreasonable.

[rebirther]

Originally Posted by VBCurtis

Are you suggesting I sieve 100-200k rather than 100-250? In the long run, this is fine, as the next sieve can be 200-500k to hit the usual next milestone. Note that base 37 is pretty low, so the tests will run quite a bit faster than bases over 100- meaning 250k isn't unreasonable.

For the smaller bases you can sieve up to 500k.

[VBCurtis]

Attached is the sieve file, with just under 42000 candidates, for base R37 from 100k to 250k. Once you finish this file, I'll sieve the remaining k's to 500k.

I sieved to 5T. This produced a time per factor of just over 400 sec, with an LLR test of the first candidate taking 125 sec on the same machine.