Sieving drive for k=2000-3400 n=50K-1M

[Svenie25]

How about individual k´s?

Lets say, you take 25 of this k´s out and say, here, this k´s are for you. Choose one or more and do it from 50K up to 1M or whereever you want to go. This includes the none-top5000-range, too.

[mdettweiler]

Originally Posted by Svenie25

How about individual k´s?

Lets say, you take 25 of this k´s out and say, here, this k´s are for you. Choose one or more and do it from 50K up to 1M or whereever you want to go. This includes the none-top5000-range, too.

I think the plan was to do all of k=2000-3400 as organized drives, but to have the option of individual k reservations for the higher k=3400-10000 range. Of course, the huge sieve for 3400-10000 won't be done for quite a while yet; it would be interesting to possibly set aside a slice of 2000-3400 for individual reservations, sort of like what we have for k=300-400.

Of course, we also want to avoid setting aside *too* large a block for individual k reservations. Although individual k's have a certain appeal, most people seem to prefer searching the team drives, for which they can find top-5000 primes much more easily.

Gary, what's your take on this?

Max

[gd_barnes]

Originally Posted by Svenie25

How about individual k´s?

Lets say, you take 25 of this k´s out and say, here, this k´s are for you. Choose one or more and do it from 50K up to 1M or whereever you want to go. This includes the none-top5000-range, too.

That is what we need to have a public discussion about. k=3000-3400 is a completely flexible range. If people like that idea, I would be in favor of pulling out about 15-20 k's for individual-k testing.

As a starting point, Karsten had previously tested k=3000-3010 and the mini-drive has been processing k=3010-3200 minus two k's in the middle. Therefore, likely the obvious choice for individual-k testing would be one of the following: k=3200-3230, k=3200-3240, k=3360-3400, or k=3370-3400.

This would allow for a much smaller (than our usual drives) main team drive of something like k=3240-3400 or k=3200-3360 in addition to the extension of k=3010-3200 in a team drive. People might like the idea of doing a team drive that can progress much faster than the huge drives that we've been doing.

What might make the most sense is to pull k=3360-3400 for individual-k testing and combine k=3010-3200 with k=3200-3360 for one larger team drive for k=3010-3360 once we get to n=500K on both smaller efforts.

Of course then we'll need to extend Karsten's k=3000-3010 or combine it in with the k=3010-3200 drive or he could re-reserve them and keep going with them.

I'm just thinking out loud right now. I like this k=3000-3400 range. It's going to be a little bit of a hassle administratively but there are a lot of different fun things that we can do with it.

Max sent me a long Email about various options on this. I haven't had time to read it in detail yet.

Likely we'll create a completely new thread for this discussion and move a few of these posts to it as it needs to be separate from the sieving drive stuff.


Gary