mini-drive for high-n testing on Sierp base 4

[gd_barnes]

David and Anon,

This sounds good to me. David, the exponent on these k/n pairs is huge, in excess of n=1.9M base 2, so the testing time on each pair is 2-3 hours. Because of this, please be aware that this server may get no activity for days or even weeks at a time. Don't be too alarmed about that. We may set up some servers like this here and there that will be dormant for long periods and then have a large spurt of activity for a while.

Anon, can you create a new thread that shows all of the current servers and their current test limits for CRUS like the one Karsten created for NPLB? I want people who already know how to run servers to have a "one stop place" where they can go and quickly know what server info. to enter into llr-clientconfig.txt and what n-range they would be testing.

To anyone interested in manual reservations for these huge tests, we can work with you. Anon and David, if someone desires a manual reservation, we can decide at that time what range to give them out of the LLRnet server. If it's before any LLRnet work has been done, it'd be easiest to give the lowest k/n pairs. After that, it will be a little trickier; perhaps the highest k/n pairs. It'd be messy to end up with 'holes' in the LLRnet k/n pairs.


Thanks,
Gary

[mdettweiler]

Originally Posted by gd_barnes

David and Anon,

This sounds good to me. David, the exponent on these k/n pairs is huge, in excess of n=1.9M base 2, so the testing time on each pair is 2-3 hours. Because of this, please be aware that this server may get no activity for days or even weeks at a time. Don't be too alarmed about that. We may set up some servers like this here and there that will be dormant for long periods and then have a large spurt of activity for a while.

Anon, can you create a new thread that shows all of the current servers and their current test limits for CRUS like the one Karsten created for NPLB? I want people who already know how to run servers to have a "one stop place" where they can go and quickly know what server info. to enter into llr-clientconfig.txt and what n-range they would be testing.

Okay, will do.

To anyone interested in manual reservations for these huge tests, we can work with you. Anon and David, if someone desires a manual reservation, we can decide at that time what range to give them out of the LLRnet server. If it's before any LLRnet work has been done, it'd be easiest to give the lowest k/n pairs. After that, it will be a little trickier; perhaps the highest k/n pairs. It'd be messy to end up with 'holes' in the LLRnet k/n pairs.

Okay, we'll see what we can do if anyone asks for manual work on this drive in the future.

Anon

[mdettweiler]

Taking 925K-1000K for LLRnet port 4.

[IronBits]

crus.ironbits.net port 4 online, plug away when ready
I set it up so that there will be a 14 day return check, per Anon's request.

[m_f_h]

Originally Posted by IronBits

crus.ironbits.net port 4 online, plug away when ready
I set it up so that there will be a 14 day return check, per Anon's request.

Hello,

I started crunching there, and noticed a possible bug in the crus:4 server's scripts : the "time" seems to be cumulated. (1st k/n pair done in 12000 sec; 2nd pair took 24000 sec)
Also, the web page's title of crus.ironbits.net
mentions the arithmetic mean of 4 and 6 ("base 5 project") - I have a slight doubt whether this is intended ;)

apart from that: nice job, running all these servers! makes it really easy for a newbie to run his client within one minute of setup (including the download of the llrnet package, if necessary).
Maybe you should put a "getting started" howto on the cited homepage ("1. get LLRnet from ..., 2. unpack in a dedicated dir, 3. put this and that in clientconfig, 4. double-click llrnet")
- or, instead of step (3), I'd suggest to provide a setup-script to double-click which patches the client_server.lua if needed (fixing the "?% complete" bug), changes the clientconfig server/port settings, asks for the username to put there, and finally starts the llrnet client.

anyway, hats off for what's already running!

[mdettweiler]

Originally Posted by m_f_h

Hello,

I started crunching there, and noticed a possible bug in the crus:4 server's scripts : the "time" seems to be cumulated. (1st k/n pair done in 12000 sec; 2nd pair took 24000 sec)
Also, the web page's title of crus.ironbits.net
mentions the arithmetic mean of 4 and 6 ("base 5 project") - I have a slight doubt whether this is intended ;)

In the results files on the server side, the timings shown are how long it took between the time the k/n pair was assigned to your client, and the time that your client returned the results. The actual timings are shown in the lresults.txt produced by your client.

As for the "base 5 project" thing in the title--whoops, looks like IronBits must have missed that when he fixed that same error in the main part of the page a while back. I'll send him an email when I get around to it (since it's not crucially important).

apart from that: nice job, running all these servers! makes it really easy for a newbie to run his client within one minute of setup (including the download of the llrnet package, if necessary).
Maybe you should put a "getting started" howto on the cited homepage ("1. get LLRnet from ..., 2. unpack in a dedicated dir, 3. put this and that in clientconfig, 4. double-click llrnet")
- or, instead of step (3), I'd suggest to provide a setup-script to double-click which patches the client_server.lua if needed (fixing the "?% complete" bug), changes the clientconfig server/port settings, asks for the username to put there, and finally starts the llrnet client.

Some instructions along those lines are available in the "Automated primality testing with LLRnet" sticky thread in this forum. The instructions are still set up for our old servers, which were hosted by Carlos (em99010pepe), but that makes for only a few minor differences between the instructions and our current setup. (I really have to get those instructions updated sometime soon. )

[em99010pepe]

As you already know I am testing the rest of the candidates left on llrnet server port 4. In a total of 256 candidates I already tested 40 % of them. I've been running them in parallel with another project but from now on they will have high priority. I hope to finish the rest until Sunday. Let's see if I can get a prime in here!

Carlos

[em99010pepe]

93 % done, no primes yet. I'll post here the results.

[em99010pepe]

Done, no primes found for k=64494 to n=2M base 2 (1M to base 4).

Input file
Results file

[gd_barnes]

Originally Posted by em99010pepe

Done, no primes found for k=64494 to n=2M base 2 (1M to base 4).

Input file
Results file

Thanks for doing that Carlos. All of Sierp base 4 is now fully complete to n=1M.

If anyone wants to take this base higher, there are 4 k's remaining: Be our guest!


Gary

[gd_barnes]

This mini-drive is now complete. The thread will be locked. Send me a PM if you think anything else should be posted here.

If and when we decide to test it above n=1M base 4, we'll create a new thread or combine reservations for it with the regular reservations threat.


Gary