PRPnet Servers for CRUS - Page 5

gd_barnes · 2015-10-17, 18:57

Thanks Peter. We're happy to have some help on port 1400. I had mostly been the only one for several months.

mdettweiler · 2016-02-17, 04:03

S17 is closing in fast on 2M in port 1300. I've had access to some extra resources lately and have been able to maintain a steady rate of n=10K/day; I expect it will be completed by the end of the month.

Any ideas on what to load next? My personal preference, as always (and keeping in the port 1300 theme), is for a base <=32 (from the "original" CRUS purview) with a good chance of near-term proof, but I'd be willing to have a go at something else if someone's got a good idea.

R23 (at n=1.1M) and S27 (at n=1.211M) are both unreserved 1-k'ers with sieve files to 2M. Are those sieves at all close to optimal, or do they need serious work before PRPing? If they're ready, I would suggest loading one or both of them into port 1300. Not familiar with the weights on these (maybe a prime in the near future isn't as realistic as it looks?

).

If they do need more sieving, does anyone have a rough estimate of how much effort would be needed to get them in shape? The bulk of my firepower right now is about 7-10 quad-core Haswells (depending on the day - I'm crunching on them in "off hours", courtesy of a script that shuts off PRPnet when someone logs on, so their uptime is somewhat variable). However, I don't have a lot of time to babysit a manual effort, so any sieving would be limited to one or two of the machines. That said, they're Haswells and thus fairly powerful (on port 1300, they're 4 times faster than my Sandy Bridge laptop! Not sure how that translates to sieving...), so they might still be able to do a lot.

VBCurtis · 2016-02-17, 05:22

If you pick R23, I'll contribute a core-month or more toward the sieve effort. Just provide timing for a single test at the low end of the file, and I can calculate a good sieve depth; I can run part of what's needed and just send you a factor file.

mdettweiler · 2016-02-17, 05:47

Sure thing. I just fired off a test for 404*23^1100204-1 (the first test in the file) on one of the Haswells; it should be done by morning.

gd_barnes · 2016-02-17, 08:08

R23 is fine with me for port 1300. Max, you and Curtis can coordinate on sieving. When the sieve is done, just send me the factors or updated sieve file.

gd_barnes · 2016-03-15, 07:05

Port 1300 has dried out. Eventually we will load R23 into it. For now, consider helping us with the various base 2/4 efforts running in port 1400.

mdettweiler · 2016-03-16, 17:44

The sieving for R23 should be done within a day or two - I'm just finishing up a couple straggling ranges.

I'll stay on port 1400 until at least n=4M, possibly farther if I can sustain my current firepower.

mdettweiler · 2016-03-20, 05:55

Curtis and I have finished sieving R23 to p=300T (which was what we estimated for optimal depth).

I'm going to hold off loading these into port 1300 for the moment, because right now all my computers are configured to run on 1300 and are just "falling back" to 1400 (and I want to stay with that a bit longer); that's easier than changing all the prpclient.ini's manually.

If anyone else is interested in starting R23 right away, please let me know and I'll load them - I don't want to hold up the project for my own convenience.

mdettweiler · 2016-04-09, 04:55

I have loaded R23, n=1.1M-2M into port 1300.

I'll continue to keep the bulk of my resources on port 1400 for now, to get that pushed higher. We should be to 4M within the week; I think I can maintain this level of firepower until at least n=4.5M. (I'd like to stay on to 5M if possible, but I may not be able to keep access to these resources for that long.)

Edit: I have also increased the test deadlines on both ports (1300 and 1400) to 1 week instead of 4 days. I think this is reasonable, considering that the tests are getting a bit longer now for both servers...at 4 days, it doesn't take much for these tests to expire, especially if you have a queue larger than 1 test, or are using computers that don't crunch full time (as is the case with most of my current resources

).

MyDogBuster · 2017-03-21, 11:39

I believe the servers are down and Gary is out of town until ???

rebirther · 2017-03-21, 17:50

Quote:

Originally Posted by MyDogBuster

I believe the servers are down and Gary is out of town until ???

Yes, the server is down and I need some files for the test on weekend. I hope all is online tomorrow?

2016-02-17, 04:03	#46
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 14151₈ Posts	S17 is closing in fast on 2M in port 1300. I've had access to some extra resources lately and have been able to maintain a steady rate of n=10K/day; I expect it will be completed by the end of the month. Any ideas on what to load next? My personal preference, as always (and keeping in the port 1300 theme), is for a base <=32 (from the "original" CRUS purview) with a good chance of near-term proof, but I'd be willing to have a go at something else if someone's got a good idea. R23 (at n=1.1M) and S27 (at n=1.211M) are both unreserved 1-k'ers with sieve files to 2M. Are those sieves at all close to optimal, or do they need serious work before PRPing? If they're ready, I would suggest loading one or both of them into port 1300. Not familiar with the weights on these (maybe a prime in the near future isn't as realistic as it looks? ). If they do need more sieving, does anyone have a rough estimate of how much effort would be needed to get them in shape? The bulk of my firepower right now is about 7-10 quad-core Haswells (depending on the day - I'm crunching on them in "off hours", courtesy of a script that shuts off PRPnet when someone logs on, so their uptime is somewhat variable). However, I don't have a lot of time to babysit a manual effort, so any sieving would be limited to one or two of the machines. That said, they're Haswells and thus fairly powerful (on port 1300, they're 4 times faster than my Sandy Bridge laptop! Not sure how that translates to sieving...), so they might still be able to do a lot. Last fiddled with by mdettweiler on 2016-02-17 at 04:08

2016-04-09, 04:55	#53
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3·2,083 Posts	I have loaded R23, n=1.1M-2M into port 1300. I'll continue to keep the bulk of my resources on port 1400 for now, to get that pushed higher. We should be to 4M within the week; I think I can maintain this level of firepower until at least n=4.5M. (I'd like to stay on to 5M if possible, but I may not be able to keep access to these resources for that long.) Edit: I have also increased the test deadlines on both ports (1300 and 1400) to 1 week instead of 4 days. I think this is reasonable, considering that the tests are getting a bit longer now for both servers...at 4 days, it doesn't take much for these tests to expire, especially if you have a queue larger than 1 test, or are using computers that don't crunch full time (as is the case with most of my current resources ). Last fiddled with by mdettweiler on 2016-04-09 at 05:07

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
PRPnet servers for NPLB	mdettweiler	No Prime Left Behind	228	2018-12-26 04:50
Public PRPNet Servers	rogue	Open Projects	26	2013-01-16 01:33
SR5 PRPnet 2.4.7 Servers - Shutting Down	Joe O	Sierpinski/Riesel Base 5	6	2010-12-06 20:41
PRPNet servers down?	opyrt	Prime Sierpinski Project	13	2009-11-04 21:33
LLRnet servers for CRUS	gd_barnes	Conjectures 'R Us	39	2008-07-15 10:26

2015-10-17, 18:57	#45
gd_barnes May 2007 Kansas; USA 101×103 Posts	Thanks Peter. We're happy to have some help on port 1400. I had mostly been the only one for several months.

2016-02-17, 05:22	#47
VBCurtis "Curtis" Feb 2005 Riverside, CA 11400₈ Posts	If you pick R23, I'll contribute a core-month or more toward the sieve effort. Just provide timing for a single test at the low end of the file, and I can calculate a good sieve depth; I can run part of what's needed and just send you a factor file.

2016-02-17, 05:47	#48
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	Sure thing. I just fired off a test for 404*23^1100204-1 (the first test in the file) on one of the Haswells; it should be done by morning.

2016-02-17, 08:08	#49
gd_barnes May 2007 Kansas; USA 101×103 Posts	R23 is fine with me for port 1300. Max, you and Curtis can coordinate on sieving. When the sieve is done, just send me the factors or updated sieve file.

2016-03-15, 07:05	#50
gd_barnes May 2007 Kansas; USA 10403₁₀ Posts	Port 1300 has dried out. Eventually we will load R23 into it. For now, consider helping us with the various base 2/4 efforts running in port 1400.

2016-03-16, 17:44	#51
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 14151₈ Posts	The sieving for R23 should be done within a day or two - I'm just finishing up a couple straggling ranges. I'll stay on port 1400 until at least n=4M, possibly farther if I can sustain my current firepower.

2016-03-20, 05:55	#52
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	Curtis and I have finished sieving R23 to p=300T (which was what we estimated for optimal depth). I'm going to hold off loading these into port 1300 for the moment, because right now all my computers are configured to run on 1300 and are just "falling back" to 1400 (and I want to stay with that a bit longer); that's easier than changing all the prpclient.ini's manually. If anyone else is interested in starting R23 right away, please let me know and I'll load them - I don't want to hold up the project for my own convenience.

2017-03-21, 11:39	#54
MyDogBuster May 2008 Wilmington, DE 2²×23×31 Posts	I believe the servers are down and Gary is out of town until ???