Team drive #1: k=400-1001 n=333.2K-600K

[gd_barnes]

Originally Posted by em99010pepe

So 36 days to complete Drive 1, can we do it?

Carlos

Of course we can! Perhaps we'll make the last rally 3 days!

[Mini-Geek]

Originally Posted by gd_barnes

Of course we can! Perhaps we'll make the last rally 3 days!

Maybe the last rally should run until Drive 1 ends. Of course, don't start it so early that the rally will last weeks, but maybe a few days. BTW so I don't slow down the ending edge of drive 1 (i.e. I'm not the last reservation to finish), do you think I should just run LLRnet after my current reservation ends or run small increments of manual files and then for the last bit run LLRnet? I'm thinking the latter, but that would mean taking new files every ~4.5 or ~9 days (either n=100 over both cores or n=100 per core) and getting the files ready, loading it into my spreadsheet to keep track of times, etc.
Will another drive start up before this one ends or not?

[gd_barnes]

Originally Posted by Mini-Geek

Maybe the last rally should run until Drive 1 ends. Of course, don't start it so early that the rally will last weeks, but maybe a few days. BTW so I don't slow down the ending edge of drive 1 (i.e. I'm not the last reservation to finish), do you think I should just run LLRnet after my current reservation ends or run small increments of manual files and then for the last bit run LLRnet? I'm thinking the latter, but that would mean taking new files every ~4.5 or ~9 days (either n=100 over both cores or n=100 per core) and getting the files ready, loading it into my spreadsheet to keep track of times, etc.
Will another drive start up before this one ends or not?

Interesting idea on the last rally. We'll consider that.

Whatever you want to do on your manual reservations is fine. Taking smaller ranges every week or so and then filling in at the end with LLRnet is no problem. You know the objective so I'm sure you won't bite off more than you can finish before year end. Thanks for asking.

On starting up a new drive before this one ends; I was personally thinking that when this one is likely down to less than a week remaining, we'll start k=400-600 for n=600K-1M for manual reservations only to start with and when this drive completes, then we'll open it up for LLRnet. At that time, we'll also open up k=600-800 and 800-1001 for n=600K-1M as well as k=1005-2000 for at least n=50K-100K if not higher. Once we get more sieving done on k=1005-2000, perhaps we'll start another drive for that k-range for top-5000 work; perhaps n=350K-500K. All of the k=1005-2000 work is still open to team discussion on how to process it.


Gary

[MyDogBuster]

On starting up a new drive before this one ends; I was personally thinking that when this one is likely down to less than a week remaining, we'll start k=400-600 for n=600K-1M for manual reservations only to start with and when this drive completes, then we'll open it up for LLRnet. At that time, we'll also open up k=600-800 and 800-1001 for n=600K-1M as well as k=1005-2000 for at least n=50K-100K if not higher. Once we get more sieving done on k=1005-2000, perhaps we'll start another drive for that k-range for top-5000 work; perhaps n=350K-500K. All of the k=1005-2000 work is still open to team discussion on how to process it.

I wouldn't mind seeing a server (maybe IB5000) doing some K work, like loading it with a couple of higher weight k's. It would help clean that out. I fear the manual k effort being left behind. People could then point a core or 2 towards that. JMHO

[gd_barnes]

Originally Posted by MyDogBuster

I wouldn't mind seeing a server (maybe IB5000) doing some K work, like loading it with a couple of higher weight k's. It would help clean that out. I fear the manual k effort being left behind. People could then point a core or 2 towards that. JMHO

You'll see me start to do much more of the individual-k work after the 1st drive is done. I'll likely do it in 'mini-drive' type fashion, where I'll reserve 5 k's at a time for n=600K-700K while others are free to reserve those same k's for n>700K. Searching above n=~900K doesn't really appeal to me very much at this point.

I'm not too concerned about them being left behind because the regular drives will have 100 k's each in them and will move along quite slowly due to the high n-ranges. If any k=300-400 gets left at n=600K for too long, we'll promote them or we (meaning me or others such as yourself with significant resources) can pick them up and move them a little higher.

I agree that k=300-400 should technically stay somewhat ahead of k=400-1001 as for how high it has been searched.

David has offered to host a private LLRnet server for anyone who wants to use one. Max is currently using one and I used one in the past.


Gary

[MyDogBuster]

Searching above n=~900K doesn't really appeal to me very much at this point.

I'm not thrilled about them either. Maybe we could take all the 900's that we (or anyone else who doesn't want to do them) reserve and put them in a private server at Davids and just point a core or 2 at them. Sort the pairs by n so that no k gets priority and have at it. I'm not sure how the output is handled, but at least we wouldn't have to monitor them. I hate to create work for others. I'm sure we could think of other things to put in that server.

[henryzz]

i think we should have a last rally on drive 1 from just before christmas(so people dont have to bother on christmas day) until new year or until the drive finishes

[IronBits]

Every day should be a drive day.
Come on over when ever you feel like it, Gary says I have plenty of pairs to go around

[gd_barnes]

Originally Posted by MyDogBuster

I'm not thrilled about them either. Maybe we could take all the 900's that we (or anyone else who doesn't want to do them) reserve and put them in a private server at Davids and just point a core or 2 at them. Sort the pairs by n so that no k gets priority and have at it. I'm not sure how the output is handled, but at least we wouldn't have to monitor them. I hate to create work for others. I'm sure we could think of other things to put in that server.

I'm assuming that you are implying to take a bunch of k's ranges for k=300-400 for n=900K-1M and put them in a server, regardless of whether those k's are reserved or are being tested for n=600K-900K. Keep in mind that when we get n-ranges thrown around with large gaps in between, it takes extra time to administer the results afterwords. It's also a little messier for Karsten to show it on his pages. The best thing to do would be for you or whomever reserves k's in the individual-k drive if you don't like n>900K is to only reserve n=600K-900K and let others who'd like to search n>900K take those. In that drive, I state that partial reservations are perfectly fine.

Why don't we discuss it more as we get closer to finishing the 1st drive? I'm still trying to sort out the mess of results from the 4th drive right now and I want to get some things moving on the CRUS project at the moment. I'm trying to avoid administering any more different efforts here at this particular moment so I can get ready for the large amount of drives we'll have here after the 1st drive is done.


Gary

[em99010pepe]

I don't mind doing all the 900's work...

[gd_barnes]

Originally Posted by em99010pepe

I don't mind doing all the 900's work...

Hum. Perhaps we can work something out.