Primegrid discussions

[MooooMoo]

Originally Posted by Rytis

A few highlights:

total credit is available on every stats page (whereas RAC is not) (please correct me if I'm wrong);
when a twin is found, the total contribution from both manual LLRing and PG (if the user moved over to PG) should be counted. LLR test count done is being recorded, and we can approximately calculate work done by multiplying M done in the original TPS by 360 (or dividing PG work done by 360, to get approximate M);
the option to see current k/n pair will be available some day.

Sounds good.

[Rytis]

For now I have switched to normal primegens to keep cruncher's computers busy while we prepare for the new n. I think it would be ideal to start working on the new n in about a week (I will only have time to modify required scripts in PG at about then), although the sieve depth is not very deep. PG removes candidates at about 2 per second, so it's faster than sieving ;)

A week would also let the current LLR queue clear, so I can archive current TPS status and restart with the new n and new empty scores without much additional work to prevent old n and new n stats mixing up. I will also create a system to allow sievers insert work directly into PrimeGrid (it will only be available to set amount of people, preferably no more than two).

I have also tested range k=2003M to k=2004M (n=195000) to see if the twin would be reported to me, and it was.

[jmblazek]

So here's a summary:

1. PG empty current queue and archive n=195000
2. PG send primegen WUs until new k for n=333333 is available
3. We get our act together and have a few M of k available in a week.

Originally Posted by Rytis

although the sieve depth is not very deep. PG removes candidates at about 2 per second, so it's faster than sieving ;)

This is good to know. However, sieving is still more "efficient" use of computing power. With that said, I see no harm in releasing several 100M of k as soon as it's available. Reason being, in about six weeks, as long as sieving continues (and we throw a few more PCs into the effort), efficiency will no longer matter. We'll probably be around 32,000,000 k left (~4,000,000 sieved & ~2,000,000 LLR'd) That's down from 38,000,000 at 205T.

What do y'all think?

Originally Posted by Rytis

I have also tested range k=2003M to k=2004M (n=195000) to see if the twin would be reported to me, and it was.

That's good to know!

[pacionet]

Is it possible for me to sieve one exponent from scratch (the next after n=333,333) and regularly upload pre-sieved files on twinprimesearch.org ?

This way, Rytis have one single point where download pre-sieved files.

I have a maximum space of 200 MB on the site, but soon I think to switch to more space.

[biwema]

You can start sieving whatever you want, if the community is interested.
For the time being, it might make more sense to help out on 333333. It needs much more effort there (if we are not that lucky).

If you start a new exponent, choose the range large enough.
100G * (yournewexp / 333333)^2 might be appropriate.

[pacionet]

Somebody told that the next exponent should be n=582,192

[KEP]

Originally Posted by pacionet

Somebody told that the next exponent should be n=582,192

I remember suggesting that, as you also pointed out, when you pointed out and referred to my "roadmap". This exponent (n=582,192) was suggested, because it will land us in the middle of the 500,000 n (which was the lower goal) and the exponent yielding a 200,000 digit prime. Thanks for using my roadmap, I think it is quite likely to have an increase in the exponents, following the increase in computing power. So my suggestion to you is that you go for n=582,192 if the voters decide to have an exponent above 500,000. Also we need a fair increase in the n exponent, otherwise our smooth project will never really make any progress, and the only thing we will do is to flood the top 5,000 primelist, but the primes wan't be noticed by other than the TPS users, so to make ourself noticed we have to follow the roadmap not exactly, but with only a few differences in n.

Regards!

KEP

Ps. How many G you should sieve I really wouldn't have a clue, but it seems that no more than 50 should be sieved to start with