CRUS LLRnet server discussion

[mdettweiler]

Originally Posted by gd_barnes

I can do that. Let me know which post #'s you want to keep or which ones you want to delete and I'll take care of it.

Actually, I was kind of thinking that we should leave this thread be, then I'd start a separate thread containing instructions on how to use the LLRnet server (i.e. the "official" announcement). Then Gary could make that a sticky. The only reason why I haven't posted it yet is because Carlos wants me to wait until he's gotten the server's power supply upgraded (it wouldn't be very good to have a server outage right after the server becomes available!).

[mdettweiler]

Originally Posted by em99010pepe

After drying the server I'll double-check myself the numbers...

Members running the LLRnet client: kar_bon and Anonymous

Okay, if you want--though, actually, I'd already doublechecked Karsten's first result and it came out fine, so I'm confident that the LLRnet server is working properly. Also, both Karsten and I are pretty sure that are systems are producing good residuals (right, Karsten? ).

Originally Posted by em99010pepe

Anon,

Your LLRNet RS5 version link is broken...

Carlos

Ooh, thanks for pointing that out. It looks like Google Pages will truncate the filename if the name is too long. I had typed the link manually, assuming that it was okay...I'll go and shorten the filenames for the SR5 ones, then post the corrected links here.

Edit: Okay, here's the links:
Windows (SR5 version)
Linux (SR5 version)

[kar_bon]

Originally Posted by Anonymous

Also, both Karsten and I are pretty sure that are systems are producing good residuals (right, Karsten? ).

why do we put some test (after completing the running Riesel-tests) where we know they must be correct. long time ago there were problems with llr (don't know exactly the version, think it was 3.3?) and they were tested by others in the old 15k-forum. i check this to find the posts and the teste, then we could put them on the server and test the results.

what do you think about this?

karsten

[mdettweiler]

Originally Posted by kar_bon

why do we put some test (after completing the running Riesel-tests) where we know they must be correct. long time ago there were problems with llr (don't know exactly the version, think it was 3.3?) and they were tested by others in the old 15k-forum. i check this to find the posts and the teste, then we could put them on the server and test the results.

what do you think about this?

karsten

Okay, that sounds like a good idea. Maybe I can grab some of the residuals for the Riesel Base 16 team drive's lower ranges, then send Carlos the file to put those numbers into the LLRnet server; we could then compare the residuals to see if anything's amiss.

[em99010pepe]

The server will be down within 45 min, I suggest to increase the wu cache to 30-40 numbers....

[mdettweiler]

Originally Posted by em99010pepe

The server will be down within 45 min, I suggest to increase the wu cache to 30-40 numbers....

Is this the aforementioned power supply upgrade?

[em99010pepe]

Originally Posted by Anonymous

Is this the aforementioned power supply upgrade?

Sorry, not yet, false alarm....

[mdettweiler]

Carlos,

Here's some numbers you can insert into the LLRnet server. They're taken from existing residuals from the lowest range of the 2nd Drive, so they'll be perfect for doublechecks. Since they're the same type of numbers that's currently in there, you won't have to run the server dry before inserting them.

All you'll need to do is paste these numbers into the end of the knpairs.txt file--without the NewPGen header, of course.

[tnerual]

Originally Posted by Anonymous

Carlos,

Here's some numbers you can insert into the LLRnet server. They're taken from existing residuals from the lowest range of the 2nd Drive, so they'll be perfect for doublechecks. Since they're the same type of numbers that's currently in there, you won't have to run the server dry before inserting them.

All you'll need to do is paste these numbers into the end of the knpairs.txt file--without the NewPGen header, of course.

first close the server ... then add the numbers then restart the server ...

if you don't do that, the server will erase the new numbers ...

for info: i think that the residues for k*2^4n-1 and for k*16^n-1 are not the same ... even if llr (classic one) reduce k*16^n-1 to k*2^4n-1

[em99010pepe]

Anon,

As tnerual said I need to stop the server before adding more work. At current speed the server will dry in less than 4 hours. (I added a fast machine)

Should I add your file? BTW, that file only has 30 numbers....I think it's a bit low....

[kar_bon]

found a test file in the old 15k-forum for testing llr version 3 which had problems with small k/n pairs. this testfile contains about 10.000 candidates from (204307539,3291) to (2232007,1490605)!
all candidates should be primes!