mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Msieve

Reply
 
Thread Tools
Old 2016-06-07, 20:45   #100
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

4,861 Posts
Default

The log doesn't list the parameters chosen; if you chose some unusual parameters, you might really need more relations. If you have a linux environment, you could try filtering with the CADO tools, though that's surely a learning curve to figure out how to convert your present data/parameters into CADO-friendly formats.

Sometimes filtering fails from too many relations; there's a flag you can invoke to tell it to use only the first, say, 90M relations. Or, you could try building with target-density=120, as that extra filtering might magically cure the "too many relations" issue. Without knowing what large-prime bounds you used, it's not obvious to me whether you have too many relations or too few. If you used lpbr/lpba = 31, 130M-150M raw relations might be necessary; but 31lp would be an ununusal choice for this size GNFS job.
I did a GNFS146 wtih 30lp, needing 75M raw relations, and I'm pretty aggressive about choosing larger large-prime bounds than most others around here (meaning I doubt any program chose 31lp on your behalf).
If 70-80M rels were all you need, using target-density might fix it; if not, you may have to use the filtering flag about number of relations ask msieve for its flag list via msieve -h).
VBCurtis is offline   Reply With Quote
Old 2016-06-08, 01:27   #101
jasonp
Tribal Bullet
 
jasonp's Avatar
 
Oct 2004

3,541 Posts
Default

It looks like the large dataset bug. There is what may be a fix in the latest SVN.
jasonp is offline   Reply With Quote
Old 2016-06-08, 11:44   #102
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

23×3×5×72 Posts
Default

Quote:
Originally Posted by jasonp View Post
It looks like the large dataset bug. There is what may be a fix in the latest SVN.
Especially if that bug is fixed it may be worth actually releasing 1.53 at some point. 1.52 is 2 years old now.
henryzz is offline   Reply With Quote
Old 2016-06-08, 12:28   #103
Alfred
 
Alfred's Avatar
 
May 2013
Germany

5×17 Posts
Default

Quote:
Originally Posted by jasonp View Post
It looks like the large dataset bug. There is what may be a fix in the latest SVN.
Thank you.


msieve 153 svn 988 is running on this c143 now.
Alfred is offline   Reply With Quote
Old 2016-06-08, 14:04   #104
Alfred
 
Alfred's Avatar
 
May 2013
Germany

5×17 Posts
Default

The same behavior.

log file is attached.
Attached Files
File Type: log 12012.2432.svn988.log (9.4 KB, 135 views)
Alfred is offline   Reply With Quote
Old 2016-06-08, 15:35   #105
Alfred
 
Alfred's Avatar
 
May 2013
Germany

5516 Posts
Default

Quote:
Originally Posted by Alfred View Post
The same behavior.

log file is attached.
I used the binary from

http://www.mersenneforum.org/showpos...5&postcount=34
Alfred is offline   Reply With Quote
Old 2016-06-08, 16:17   #106
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

250018 Posts
Default

Could a case be made for DEFAULT_TARGET_DENSITY being set to something higher than 70.0?
xilman is offline   Reply With Quote
Old 2016-06-08, 16:25   #107
Alfred
 
Alfred's Avatar
 
May 2013
Germany

5·17 Posts
Default

TD was set to 120 (in the run with msieve 1.53 svn 988)
Alfred is offline   Reply With Quote
Old 2016-06-08, 18:27   #108
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

10,753 Posts
Default

Quote:
Originally Posted by Alfred View Post
TD was set to 120 (in the run with msieve 1.53 svn 988)
We may be talking at cross porpoises. In the source code here:

Code:
pcl@anubis ~/nums/msieve-code/trunk $ svn update
Updating '.':
At revision 993.
pcl@anubis ~/nums/msieve-code/trunk $ grep DEFAULT_TARGET_DENSITY */*/*
grep: b40c/b40c/radix_sort: Is a directory
grep: b40c/b40c/util: Is a directory
grep: b40c/test/bin: Is a directory
grep: build.cuda.vc12/gnfs/gnfs: Is a directory
grep: build.cuda.vc12/msieve/gnfs: Is a directory
grep: build.cuda.vc12/sort_engine_sm20/b40c: Is a directory
grep: build.cuda.vc12/stage1_core_sm20/gnfs: Is a directory
grep: cub/cub/agent: Is a directory
grep: cub/cub/block: Is a directory
grep: cub/cub/device: Is a directory
grep: cub/cub/grid: Is a directory
grep: cub/cub/host: Is a directory
grep: cub/cub/iterator: Is a directory
grep: cub/cub/thread: Is a directory
grep: cub/cub/warp: Is a directory
gnfs/filter/filter.c:#define DEFAULT_TARGET_DENSITY 70.0
gnfs/filter/filter.c:	merge->target_density = DEFAULT_TARGET_DENSITY;
grep: gnfs/poly/stage1: Is a directory
grep: gnfs/poly/stage2: Is a directory
pcl@anubis ~/nums/msieve-code/trunk $
the default target density is 70.0. If you want it to be 120 then you must set it explicitly on the command line.

Personally I'd put the #define in filter.h but that's a different and entirely stylistic matter.
xilman is offline   Reply With Quote
Old 2016-06-08, 18:39   #109
Alfred
 
Alfred's Avatar
 
May 2013
Germany

5×17 Posts
Default

Code:
Wed Jun 08 14:04:15 2016  setting target matrix density to 120.0
The log file contains that line, so TD=120 is valid.
Alfred is offline   Reply With Quote
Old 2016-06-08, 20:47   #110
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

10,753 Posts
Default

Quote:
Originally Posted by Alfred View Post
Code:
Wed Jun 08 14:04:15 2016  setting target matrix density to 120.0
The log file contains that line, so TD=120 is valid.
Yes, we are talking at cross porpoises. You're referring to one particular run where the target density was specifically set to a (non-default) value of 120. I'm asking whether the default for all runs should be set higher than 70, which is what you get if you don't specify a value.
xilman is offline   Reply With Quote
Reply



Similar Threads
Thread Thread Starter Forum Replies Last Post
Msieve 1.50 feedback firejuggler Msieve 99 2013-02-17 11:53
Msieve v1.48 feedback Jeff Gilchrist Msieve 48 2011-06-10 18:18
Msieve 1.43 feedback Jeff Gilchrist Msieve 47 2009-11-24 15:53
Msieve 1.42 feedback Andi47 Msieve 167 2009-10-18 19:37
Msieve 1.41 Feedback Batalov Msieve 130 2009-06-09 16:01

All times are UTC. The time now is 01:09.


Sat Jul 17 01:09:19 UTC 2021 up 49 days, 22:56, 1 user, load averages: 1.10, 1.60, 1.53

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.