[QUOTE=AMDave;159348]Found a bug! Good man.
I squished it.
Catch up is in progress.[/QUOTE]

Catch up complete.
Normal running resumed.

[QUOTE=gd_barnes;159359]Port 9000 is the only one where we should not be getting Email notification on. All others, we should.

I've been faithfully getting all of the ones for port 8000.[/QUOTE]

How about a Prime Score threshold across all ports (like => 0.5)?

At what point does the value / weight / score of the prime become of interest to you?

While you are thinking about that, try sorting the NPLB Primes List by score
(click on the Prime Score column heading)
It should look like this: [url]http://stats.ironbits.net/statsnew/prime_list.php?order=prime_score&type=desc[/url]
PS - Well done Max!

[quote=AMDave;159362]How about a Prime Score threshold across all ports (like => 0.5)?

At what point does the value / weight / score of the prime become of interest to you?

While you are thinking about that, try sorting the NPLB Primes List by score
(click on the Prime Score column heading)
It should look like this: [URL]http://stats.ironbits.net/statsnew/prime_list.php?order=prime_score&type=desc[/URL]
PS - Well done Max![/quote]
Well, it's not so much the exact size of the primes that we're concerned about regarding which ones to report; it's mainly the n-value. If n is less than or equal to 200,000, then it shouldn't be reported, since Gary will just get all the primes from the results; however, if n is greater than 200,000, then an email notification should be sent since the prime will need to be reported in the respective primes thread here in the forum. :smile:

[quote=mdettweiler;159375]Well, it's not so much the exact size of the primes that we're concerned about regarding which ones to report; it's mainly the n-value. If n is less than or equal to 200,000, then it shouldn't be reported, since Gary will just get all the primes from the results; however, if n is greater than 200,000, then an email notification should be sent since the prime will need to be reported in the respective primes thread here in the forum. :smile:[/quote]

Not necessarily. It is on a case-by-case basis. In this case, it's probably best to leave port 9000 as a no-Email port, even for the range of n=200K-350K. So the bottom line for you server guys: Continue to send no Emails on port 9000 unless you hear otherwise.

I'm saying that because it will be more accurate if the admins continue to report the primes in groupings instead of having people report them individually, although it can be done in k=25-50 groupings for this higher n-range. It shouldn't be near the effort with far fewer primes in the higher range. With this go around, we can also report them separately in the primes thread in addition to the 1st post of each drive. This will work well because to keep it easier on Karsten, we'll continue to search by k for n=200K-350K.

On another note, Max, can you please process a bunch of results, both here and at CRUS? You might also check the 9th drive and see if all stragglers are done for k=1600-1800. I'd like to get the primes checked off on as many as possible before our n=352K-360K range is done on the 8th drive. I asked because it'll only be another 1-2 days before that range is done. At that point, I'll pull down a few quads off of it and spread them out a little. Combine that with the 9th drive finishing n=50K-200K in < 3 days will take the pressure off of us admins quite a bit. I think you definitely want to have as much processed as possible before we start the rally.


Gary

[QUOTE=gd_barnes;159470]So the bottom line for you server guys: Continue to send no Emails on port 9000 unless you hear otherwise.[/QUOTE]

Roger that.
Out.

[quote=henryzz;158974]why couldnt we have just started the server up again on one of gary's or david's machines[/quote]

[quote=mdettweiler;158976]You're right, we could have done that easily. However, it looks like Ian's already willing to do the work, so thus we didn't bother setting up a server for it.

However, we *will* be using the now-idle G8000 LLRnet server for the upcoming individual-k drive (which includes k=315, formerly loaded into C6000). We'll be starting that effort sometime later this month after the sieving drive has completed and we've restarted the 8th Drive.[/quote]


Henry or anyone else,

Since the server running k=341 effectively got stopped right in the middle of processing with the remaining pairs handed off to Ian, I was thinking that we could instead load k=319 into my port G8000 and hack away at it for a while. Since it's somewhat higher-weight than k=341, I'd just have Max load it in n=100K at a time starting with n=600K-700K.

After we are done with k=319 or we determine it's time to move on, we can then start our 6k mini-drive as previously mentioned in the individual-k thread by loading it into port G8000 after straggling pairs have cleared out for k=319.

We'd be using k=319 since it's the lowest k at NPLB with no work having been done on it yet.


Gary

Is this even possible? Look at the top two from Max.
[url]http://stats.ironbits.net/statsnew/prime_list.php?order=prime_score&type=asc[/url]

[quote=IronBits;159801]Is this even possible? Look at the top two from Max.
[URL]http://stats.ironbits.net/statsnew/prime_list.php?order=prime_score&type=asc[/URL][/quote]


Those are a couple of personal results from Max. I'm not sure those kinds of efforts should be counted in our total stats but they are likely to be laying around here and there so I wouldn't mess with it.

Yes, it is possible. The k can be as large as you want it to. I've seen k's > 40 digits before.


Gary

[QUOTE=gd_barnes;159802]Those are a couple of personal results from Max. I'm not sure those kinds of efforts should be counted in our total stats but they are likely to be laying around here and there so I wouldn't mess with it.
[/QUOTE]

i think those pairs/primes should be deleted because they distort the primes found by NPLB on regular efforts!
i did so for my personel stats for checking all LLRnet-results!

David,

There is a big pile of invalid lines in the joblist.txt file for port 8000. It appears to be 4025 "jobs" that stem from the start of the n=200K-210K range back on Jan. 2nd that was abrubtly halted. They don't seem to be affecting anything but they do make it a little inconvienient to see who has the first true k/n pair in their queue for an extended period.

Let's consult with Max first but do you think it would be a problem to delete them before the rally starts? I have attached the exact lines that need to be deleted. They all have a k-value in the n=200000-210000 range. You can just do a find on "/3" to get to the first good job and delete all of the jobs above it being careful not to delete any extra commas or brackets. (sorry if I'm stating the totally obvious here)


Max,

Do you see any undue risk from this? I believe David would have to temporarily stop the server, carefully delete the bad lines, and then restart the server. Do I understand that right?


Gary

David,

I have now checked primes.txt for port 9000 up to k=1800 vs. the results files. Everything is matching up but there is a double entry for one prime in there:

1773 181608 IronBits
[2009-01-18 19:54:42]
1773*2^181608-1 is prime! Time : 65.0 sec.
1773 181608 IronBits
[2009-01-18 19:56:23]
1773*2^181608-1 is prime! Time : 166.0 sec.


Were there some pairs that got processed twice? There shouldn't have been because Max would have let me know if the results didn't match the file that we sent you for this range.


Gary