100M L-L time limit

Chuck · 2011-06-07, 20:06

Just for fun I requested a 100M digit L-L test for one of my three workers (using two CPU threads for each worker on a six-core I7-970. The estimated time to completion is greater than one year and a warning showed up in the communications window that the exponent coud be reassigned to another user after one year.

If I make continuous progress on this large exponent, is there really any danger that it could be reassigned? I wouldn't want to waste a year of processing time.

Chuck

Prime95 · 2011-06-07, 22:36

It will not be reassigned as long as your computer checks in at least once every 60 days or so.

I suggest timing a few thousand iterations to see if you have the stomach to wait for the result. It could be a few years.

Chuck · 2011-06-07, 22:49

I did check the timing — it estimates 450 days with four threads working, so I think I will probably reduce the other two workers to one thread each and give this one four.

Chuck

Christenson · 2011-06-08, 02:40

Hqve we done deep TF and P-1 on these 100M exponents?

KingKurly · 2011-06-08, 03:16

Quote:

Originally Posted by Christenson

Hqve we done deep TF and P-1 on these 100M exponents?

See http://mersenneforum.org/showthread.php?t=10693&page=12

I can't speak for Chuck's exponent, but yes, there has been TF and P-1 work done in that range. I did one of the P-1 tests: at the time that it finished back in April, it was the 97th exponent in the range to be P-1'd.

davieddy · 2011-06-08, 06:33

Quote:

Originally Posted by Prime95

I suggest timing a few thousand iterations to see if you have the stomach to wait for the result. It could be a few years.

Also glance at the improbability of it being prime!

What is so special about base ten?*

...but if it floats your boat, Carry On Hunting.

David

*Ask Paul about Norfolk jokes.

Prime95 · 2011-06-08, 13:28

Quote:

Originally Posted by Chuck

I did check the timing — it estimates 450 days with four threads working, so I think I will probably reduce the other two workers to one thread each and give this one four.

Doublecheck those timings using the wall clock. The Test/Status dialog box is not great at estimating when using multi-threaded FFTs.

2011-06-07, 20:06	#1
Chuck May 2011 Orange Park, FL 3·5·59 Posts	100M L-L time limit Just for fun I requested a 100M digit L-L test for one of my three workers (using two CPU threads for each worker on a six-core I7-970. The estimated time to completion is greater than one year and a warning showed up in the communications window that the exponent coud be reassigned to another user after one year. If I make continuous progress on this large exponent, is there really any danger that it could be reassigned? I wouldn't want to waste a year of processing time. Chuck

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Limit on TF assignments?	Rodrigo	PrimeNet	21	2016-05-01 15:42
CPU time for 100M digit prime test	biggerben	Software	7	2014-10-24 05:47
GNFS poly search time limit oddity	Andi47	Msieve	13	2009-02-25 12:33
Is it time for a 100M digit option?	petrw1	PrimeNet	60	2008-09-30 22:43
Has the editing time limit been changed?	jasong	Forum Feedback	2	2006-06-22 20:00

2011-06-07, 22:36	#2
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 1110101100110₂ Posts	It will not be reassigned as long as your computer checks in at least once every 60 days or so. I suggest timing a few thousand iterations to see if you have the stomach to wait for the result. It could be a few years.

2011-06-07, 22:49	#3
Chuck May 2011 Orange Park, FL 1565₈ Posts	I did check the timing — it estimates 450 days with four threads working, so I think I will probably reduce the other two workers to one thread each and give this one four. Chuck

2011-06-08, 02:40	#4
Christenson Dec 2010 Monticello 703₁₆ Posts	Hqve we done deep TF and P-1 on these 100M exponents?