Rally Jan. 23rd-25th

[gd_barnes]

Originally Posted by em99010pepe

No thanks, I am out. I'll will stay on the 9th drive.

Famous last words. HA-HA-HA!!!

[IronBits]

Beyond has yet to show up with the rest of his hungry mutts...
Bok has not had time to add all his computers yet either...

Glutton for punishment are we? :p
You guys are bleeding to death...

Also note, I said *almost* based on the 1st hours report.

I need to stoke up that BBQ fire a tad more I see. ;)

[gd_barnes]

Originally Posted by AMDave

Cerrected deficit based on last hour's output is now 52,354 pairs short of the 48 hours.

The tests will slow down a little bit as the n-value gets higher. But...I suspect we'll still be short.

David, late tonight or sometime or Saturday, go ahead and load up the n=400K-420K pairs that I sent you last night.

Impressive performance by Free-DC so far. Can you maintain even double the ROLP output? Down to 2.4X our pairs now. You better not let up...38 hours still to go!


Gary

[IronBits]

Server is not working hard at all.
Compared to port 9000 scrolling on by, I'd say we could handle lots and lots more.
No more worries on future rallies I guess <yawn>

[gd_barnes]

Originally Posted by IronBits

Server is not working hard at all.
Compared to port 9000 scrolling on by, I'd say we could handle lots and lots more.
No more worries on future rallies I guess <yawn>

...amazing servers to say the least, David. Nice work! They appear to be fully professional grade; that is able to handle large #'s of corporate clients.

I estimated that just to get to the peak load of port 9000, we'd have to have 1600 cores on port 8000. And...as I recall, port 9000 could handle far more than it's highest point, because I believe you said it hadn't exceeded socket 17-18 out of 125 sockets or so.

That tells me that you could handle upwards of 10,000 cores (conservative estimate) on port 8000 at the current testing rate.




Gary

[gd_barnes]

One more thing: I think it is great to have a server that can handle the super-fast tests like port 9000. Those small tests are just a nuisance but are highly necessary for NPLB to keep up the montra of No Prime Left Behind and it's great for folks who love to see how many gillions of results they can process in a day, week, or month! Even I couldn't resist putting 7 of my 10 quads on it for a day-and-a-half just to see what I could do.

We could even consider tests as small as n=10000 or 20000 in the future. Psst...don't tell Karsten though. It's enough primes for him to keep up with starting from n=50000.


Gary

[IronBits]

Glad you approve of our progress and am very happy that it's turned out so well. You may not have realized this, but while we were going hard on port 9000, it was still handling all the other ports you have running at that same time.
Plus, I downloaded CentOS 5.2 for Bok's new stats server, a large demo game and Windows 7 in the background.

AMDave has been working tirelessly in the background adding new features and adding enhancements. Soon the auto-notify will work and everyone will start getting double notifications... once we are happy with it, or you all get tired of duplicate emails, I'll stop my scripts and then there will be only one thing left to do before the total conversion is complete.
That status web page I create so you can keep an eye on what is going on with each port is the last thing to be done.

Did you see the new admin panel AMDave created so you can turn on/off email notifications by port and Server.
He has written it so that if it finds a new port it wasn't aware of, it will just add it into the database with the default of Email=Y on that port number and server
Then you guys can toggle it on or off as need be...

[IronBits]

Let us know of anything else you would like to have, or would be nice to have...
How about something to research ranges so you can try to find missing pairs or something?

[gd_barnes]

Originally Posted by IronBits

Glad you approve of our progress and am very happy that it's turned out so well.

AMDave has been working tirelessly in the background adding new features and adding enhancements. Soon the auto-notify will work and everyone will start getting double notifications... once we are happy with it, or you all get tired of duplicate emails, I'll stop my scripts and then there will be only one thing left to do before the total conversion is complete.
That status web page I create so you can keep an eye on what is going on with each port is the last thing to be done.

Did you see the new admin panel AMDave created so you can turn on/off email notifications by port and Server.
He has written it so that if it finds a new port it wasn't aware of, it will just add it into the database with the default of Email=Y on that port number and server
Then you guys can toggle it on or off as need be...

More excellent features! That will be helpful! Thanks!

Along those lines: I've spent the last 1-2 hours checking all primes found since 12:01 AM on Jan. 22nd; 14 hours before the rally started. I have checked the following:

1. Emails sent out.
2. Primes listed on your web page in the 1st link of the 1st post of this thread.
3. Primes listed in our "k>1001 primes" thread.
4. Primes listed in the 1st post of our 8th drive thread.
5. Primes on the top-5000 site.

What I found was that there are 8 primes where no Email was sent between 7:30 AM and 11:53 AM this morning. Here is a breakdown of who found the primes where no Email was generated:

Beyond 4
PCZ 2
IronBits 1
LaurenU2 1

I went back through my listing and I see that Beyond found no other primes. But PCZ, LaurenU2, and you have all had other primes where Emails were generated.

Interestingly, they all occurred in that one 4-1/2 hour timeframe. Before that and since then, Emails have been received on all of them. Keep in mind that I can only count them if I got copied in on them. I'm assuming that you copied Max and I in on everyone's primes.

Can you check into that?


Gary

[gd_barnes]

David,

In doing the verification in the above post, I have also found the following:

For reporting top-5000 primes, it's been better than expected for as many new people that we have running the rally. Excluding the last 2 hours where 3 new primes have been found but the people aren't around at the moment, there are only 2 new primes that have not been submitted to top-5000.

Both new primes are by LaurenU2. He also has multiple confirmed primes that I'll go ahead and list in our primes thread.

Does LaurenU2 know anything about submitting top-5000 primes? If not, does he want to? If not, I think Ian suggested that we create some misc. "anonymous NPLB ID" at top-5000 to get them submitted; something that I agree with.

We appreciate the huge crunching power of whomever enjoys running rallies with us but we just need to know how to handle these things that they likely know little about and may not be concerned about.

David, Carlos, James, or other Free-DCers; any thoughts on this? LaurenU2 has received Email notification of the primes and an Email from me pointing him to the thread for new people for the rally. I'll follow up again with more specifics on reporting primes shortly and see if I get a response by mid-day Sat.


Thanks,
Gary

[em99010pepe]

Originally Posted by IronBits

I spoke with Lauren and he tells me he is only running 1 client per core.
So I don't know why there are so many rejected from him.

Anyone else have any clues?

I have....you are using the sr5 llrnet version. Maybe that's the issue. Here's my llrnet win version:

http://em99010pepe.googlepages.com/llrnet.zip

You can see that it is 2x the size of the one you use. Maybe there was an error on compiling the sr5 llrnet version.

Edit: the sr5 llrnet version doesn't have the GUI option.

Carlos