Thoughts on future NPLB searches

[em99010pepe]

MrOzzy,

As you already know we are almost finishing Drive 1 and because you are the only dedicated member using C443 server (more two members helping just until the end of the drive) I would like to know what type of work would you be interested to have on the server.

[MrOzzy]

You know you can send me a PM too :-).
Anything really. I'm not sure what options I have here, but for example searching lower n ranges is fun because you can find lots of primes quickly. On the other hand searching one or more fixed k for a large n range can be quite interesting too ...
Or fill up a gap someware ..

[em99010pepe]

I know, just wanted to have a public discussion with you. I am interested in cleaning the k's from this thread, your second option.

[MrOzzy]

Oh sure, go ahead. Work your way down the list I would say, or from bottom to top, or whatever you'd like, all seems fine for me.

[gd_barnes]

I renamed this thread because it is a good starting point on discussing thoughts on future NPLB searches.

Carlos, this was a good idea to publicly start a discussion about future searches here.

One thing that I'm considering: Allow searching of any n-range in the k=300-400 individual-k drive. It wouldn't have to be contiguous. We do need to avoid any situations where there would be more than one gap on any one k. So for a k already searched to n=600K, it would be bad to search n=700K-750K and 800K-850K but it would be fine to only search n=900K-1M. That would free you up to search multiple k's at higher n-ranges.

Max and Karsten, do you feel that this would be manageable administratively? I don't think it would be too difficult to show 2 different search ranges with a single gap on various k's for k=300-400.

Another thing that I thought of: Have a "choose your own k" effort for k>2000 like RPS does but do it a little differently. Wait until there are at least 5-10 random k's that people would like to search and then do a semi-coordinated team sieve for them. That would be far more efficient than several people sieving 1-2 k's at once.

If we do any kind of changes of this nature, I'd prefer that we wait until Jan. after the new drives that we will have are already kicked off.

If anyone has any other thoughts, please post them here.


Gary

[em99010pepe]

It's going to be a mess if we don't search continuously for each k.
What about starting a new drive where we choose the k's (300<k<400)? I would pick up the hight weigh k's like 315 (I'll free up this one from my reservation), 321, 339, 357, 363 and 375.

Carlos

[gd_barnes]

Originally Posted by em99010pepe

It's going to be a mess if we don't search continuously for each k.
What about starting a new drive where we choose the k's (300<k<400)? I would pick up the hight weigh k's like 315 (I'll free up this one from my reservation), 321, 339, 357, 363 and 375.

Carlos

I agree it would be a bit messy but am willing to accomodate people to a certain extent if they want to do higher-n searches. That said, I like the idea of your drive of heavy-weight k's much better. k=315 would take a long time to search all by itself by one person.

I think we will go with this as long as Max and Karsten are good with it and we have some further discussion about which k's and how many of them to include. What I will suggest doing is calling them "mini-drives" and not make them "formal" drives so to speak like our k=400-1001 for n=600K-1M and k=1005-2000 for n=50K-500K drives will be. In other words, we won't sticky them and will just call them something like "k=300-400 mini drive A", "k=300-400 mini drive B", etc.

The good thing about such drives is that it will allow people to search for higher primes much more quickly without having to worry about completing an entire k instead of having to wait until our main drives get there, which will take quite a bit of time.


Gary

[kar_bon]

i got no problems with displaying 2 searched ranges on the k<300-page then.

for such efforts the results will collected by Gary (hope so), so we can be sure all ranges will be done at the end (not like k=31 today!!! the searched range was given to n=1.3M but there's a prime missing at n=406105!!!!!! how could this come when all results were collected?)

[henryzz]

Originally Posted by kar_bon

i got no problems with displaying 2 searched ranges on the k<300-page then.

for such efforts the results will collected by Gary (hope so), so we can be sure all ranges will be done at the end (not like k=31 today!!! the searched range was given to n=1.3M but there's a prime missing at n=406105!!!!!! how could this come when all results were collected?)

we dont want something like this either
http://www.mersenneforum.org/showpost.php?p=151090&postcount=3

[mdettweiler]

Originally Posted by gd_barnes

I agree it would be a bit messy but am willing to accomodate people to a certain extent if they want to do higher-n searches. That said, I like the idea of your drive of heavy-weight k's much better. k=315 would take a long time to search all by itself by one person.

I think we will go with this as long as Max and Karsten are good with it and we have some further discussion about which k's and how many of them to include. What I will suggest doing is calling them "mini-drives" and not make them "formal" drives so to speak like our k=400-1001 for n=600K-1M and k=1005-2000 for n=50K-500K drives will be. In other words, we won't sticky them and will just call them something like "k=300-400 mini drive A", "k=300-400 mini drive B", etc.

The good thing about such drives is that it will allow people to search for higher primes much more quickly without having to worry about completing an entire k instead of having to wait until our main drives get there, which will take quite a bit of time.


Gary

Okay, sounds good to me.

[masser]

Originally Posted by henryzz

we dont want something like this either
http://www.mersenneforum.org/showpost.php?p=151090&postcount=3

Umm... I followed the links to the original post by ltd - first of all, the quote is taken out of the context of the original thread it was in - was this a thread that was occasionally purged of older content? was it a thread where the original readers and posters knew the background info about ranges previously searched? Second, I believe ltd is a non-native english speaker, so his saying "no primes found" might easily be considered/interpreted as "no new primes found" It's a little harsh to call that post (missing one adjective that when taken out of context can look erroneous) the Most Erroneous post. Certainly there are some good candidates for that title in the Miscellaneous threads, and I would venture that there are some good candidates in the Soapbox thread as well.