mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Closed Thread
 
Thread Tools
Old 2009-04-09, 00:04   #78
Flatlander
I quite division it
 
Flatlander's Avatar
 
"Chris"
Feb 2005
England

31·67 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
...
Chris, just to confirm...you'll be finishing your range sometime on Apr. 9th, correct?
...
Gary
Yes, late afternoon GMT. (Unless I get another power cut...)
Flatlander is offline  
Old 2009-04-09, 00:26   #79
Brucifer
 
Brucifer's Avatar
 
Dec 2005

4718 Posts
Default

Gary I just answered your email. I split that range over two Q9400's, 8 cores. Around 9:30am PST on the 10th is when my part of the 20T will be finished except for the range you loaded on to your systems. I hope that works for you. See the email please... the one labeled the crises. :)

Last fiddled with by Brucifer on 2009-04-09 at 00:27
Brucifer is offline  
Old 2009-04-09, 00:33   #80
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

240538 Posts
Default

Quote:
Originally Posted by Brucifer View Post
Gary I just answered your email. I split that range over two Q9400's, 8 cores. Around 9:30am PST on the 10th is when my part of the 20T will be finished except for the range you loaded on to your systems. I hope that works for you. See the email please... the one labeled the crises. :)
Excellent! I hadn't realized you could start on it right away and had assumed that you would have to do 25G per core instead of the 20G that you stated in your Email. I'll change your ETA to 4/10 in the AM for 19850e9-2000e9.

David, it's not so bad now. The sieving should be done to P=20T by 6-8 PM EDT on Apr. 10th. Port 8000 wouldn't dry, at its current rate, until close to noon on Apr. 11th so we're in decent shape. I'll keep an eye on its testing rate. If someone adds a bunch of cores, I'll pull some of mine over to port 5000 for a while.


Gary

Last fiddled with by gd_barnes on 2009-04-09 at 00:35
gd_barnes is offline  
Old 2009-04-09, 00:44   #81
Brucifer
 
Brucifer's Avatar
 
Dec 2005

313 Posts
Default

Dumb question time...... Looking down the road on the rally, is any of this stuff going to be used? If so, how much more needs to be completed. Just asking before I reload stuff after this 20T stuff is finished.
Brucifer is offline  
Old 2009-04-09, 01:20   #82
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

7×13×113 Posts
Default

Quote:
Originally Posted by Brucifer View Post
Dumb question time...... Looking down the road on the rally, is any of this stuff going to be used? If so, how much more needs to be completed. Just asking before I reload stuff after this 20T stuff is finished.
I'm a little confused by your question but will answer what I think you're after:

Once we hit P=20T (20000e9), we'll use those factors for the n=520K-600K range files of the 10th drive and so it will be quite a while before we need any more factors. The rally will not come even close to getting up to n=600K even if we have the same # of resources as the last huge rally. (Not likely because it appears that Beyond is no longer available.) I'm hoping we'll near n=550K by the end of the rally, which would be a substantial accomplishment over the next 10 days or so, but n=540K would be very good also.

Therefore, you can take your time with 20T-30T because we won't use those factors until the 10th drive is nearing n=600K. I would suggest loading something like P=250G ranges into each core so you aren't messing with them stopping and having to restart new ranges frequently. Also, it's easier for me to handle fewer large factor files than many smaller files.

We're leaving the entire n-range in the file and continuing to sieve it to P=30T because it takes little additional time to sieve n=500K-1M vs. 600K-1M. We will use the deeper sieved file for double-checking in the future, which will save time there.

To go math-geek on you: Sieving n=500K-1M vs. 600K-1M takes sqrt(500K/400K) longer = 1.118 = 11.8% more time and will save us more than that on double-checking in the future.

Future double-checking is also why we have the k=1003-1400 range in there even though we're not currently testing it. But there's another reason: It's possible that Benson may not want to take k=1003-1400 all the way up to n=1M (he's stated in an Email that he is taking it to n>600K but hasn't specified how high yet) so we will be here to pick up where he left off if he stops.

In other words, we're trying to cover all of the possibilities in the future by continuing to sieve the entire k and n-range.

Does that answer what you are after or way more than you're after?


Gary

Last fiddled with by gd_barnes on 2009-04-09 at 01:23
gd_barnes is offline  
Old 2009-04-09, 02:52   #83
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

7·13·113 Posts
Default

Ah, we're in even better shape than I thought. My prior estimate was extrapolated from when I was sieving far lower P-ranges. With far less pairs in the file now, the sieving is faster. Right from the source itself, sr2sieve, my sieving range should complete about 7 AM Friday morning CDT.

David, we'll have more than a day leeway on the next file for port 8000 now.
gd_barnes is offline  
Old 2009-04-09, 06:46   #84
Brucifer
 
Brucifer's Avatar
 
Dec 2005

4718 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Does that answer what you are after or way more than you're after?

Gary
yup. :)
Brucifer is offline  
Old 2009-04-09, 14:20   #85
Flatlander
I quite division it
 
Flatlander's Avatar
 
"Chris"
Feb 2005
England

31×67 Posts
Default

17-17.25T complete.
Quote:
Found factors for 5355 terms in 1294896.026 sec. (expected about 5358.63)
Attached Files
File Type: zip factors.zip (65.8 KB, 72 views)
Flatlander is offline  
Old 2009-04-10, 01:05   #86
Brucifer
 
Brucifer's Avatar
 
Dec 2005

313 Posts
Default

Yo Gary... the faster one is now averaging 224000 p/sec per core. The slowest is doing 203/204 so big change. I wasn't using v1.7.15, so updated them. Changes the projections by days like from a completion on 1 May down to 22 April for a 250G chunk per core. So they are running like yours now. :)
Brucifer is offline  
Old 2009-04-10, 04:41   #87
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

7·13·113 Posts
Default

Quote:
Originally Posted by Brucifer View Post
Yo Gary... the faster one is now averaging 224000 p/sec per core. The slowest is doing 203/204 so big change. I wasn't using v1.7.15, so updated them. Changes the projections by days like from a completion on 1 May down to 22 April for a 250G chunk per core. So they are running like yours now. :)

Excellent! I'm glad we checked your P-rate. Once I get all the factors up to P=20T on Friday, I'll compute the optimum sieve depth and send you a new file.

For everyone's reference: It's looking like we'll probably need to sieve this range to P=40T-50T somewhere. A bad error on the original estimate on my part. With both the k-range and n-range being larger than any previous sieving effort, there is quite a bit to be gained in total testing time to sieve deeper than other efforts. Based on that, I'll probably open back up the sieving drive here. As for me, after the end of the rally, I'll probably keep one full quad on it full time until it's done.

One final note: Bruce reported completion of P=18750G-19750G in an Email.


Gary

Last fiddled with by gd_barnes on 2009-04-10 at 04:59
gd_barnes is offline  
Old 2009-04-10, 04:58   #88
IronBits
I ♥ BOINC!
 
IronBits's Avatar
 
Oct 2002
Glendale, AZ. (USA)

3·7·53 Posts
Default

If I were to take P=40-50T range...
If you break it up to go across 4 x 8 cores = 32 * 3.5 GHz cores
How long would that take?
IronBits is offline  
Closed Thread

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Team drive #10 k=1400-2000 n=500K-1M gd_barnes No Prime Left Behind 61 2013-01-30 16:08
Misc. reserves/statuses/primes for k>2000 & k=1003 gd_barnes No Prime Left Behind 113 2009-07-30 22:32
Sieving drive for k=2000-3400 n=50K-1M gd_barnes No Prime Left Behind 145 2009-06-23 18:28
Team drive #8 k=1400-2000 n=350K-500K gd_barnes No Prime Left Behind 101 2009-04-08 02:11
Sieving drive for k=1005-2000 n=200K-500K gd_barnes No Prime Left Behind 118 2009-01-17 16:05

All times are UTC. The time now is 22:26.

Sun Jan 24 22:26:00 UTC 2021 up 52 days, 18:37, 0 users, load averages: 1.31, 1.85, 1.95

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.