Yafu - Page 74

Andi_HB · 2011-08-23, 16:31

I have just got an error while i am running the script from Yoyo to do some work for the DB.
I have checked the factor.txt output and it looks like it`s a problem from Yafu. Yafu shutdown 2x after random seeds.
I have used the Yafu-32k-x64 Version on Core i3 with win7.
Till now it was running without problems (heat....).
Can someone please try this number on 64bit system.
http://www.factordb.com/index.php?sh...00000452281216

Code:

08/23/11 18:14:15 v1.28.3 @ LAPPIE, Starting factorization of 7205584074334421341608887546232682732704425079530517019873416769
08/23/11 18:14:15 v1.28.3 @ LAPPIE, ****************************
08/23/11 18:14:15 v1.28.3 @ LAPPIE, rho: x^2 + 1, starting 1000 iterations on C64
08/23/11 18:14:15 v1.28.3 @ LAPPIE, rho: x^2 + 3, starting 1000 iterations on C64
08/23/11 18:14:15 v1.28.3 @ LAPPIE, rho: x^2 + 2, starting 1000 iterations on C64
08/23/11 18:14:15 v1.28.3 @ LAPPIE, pp1: starting B1 = 20K, B2 = gmp-ecm default on C64
08/23/11 18:14:15 v1.28.3 @ LAPPIE, pp1: starting B1 = 20K, B2 = gmp-ecm default on C64
08/23/11 18:14:15 v1.28.3 @ LAPPIE, pp1: starting B1 = 20K, B2 = gmp-ecm default on C64
08/23/11 18:14:16 v1.28.3 @ LAPPIE, pm1: starting B1 = 100K, B2 = gmp-ecm default on C64
08/23/11 18:14:16 v1.28.3 @ LAPPIE, Finished 25 curves using Lenstra ECM method on C64 input, B1 = 2K, B2 = gmp-ecm default
08/23/11 18:14:16 v1.28.3 @ LAPPIE, Finished 9 curves using Lenstra ECM method on C64 input, B1 = 11K, B2 = gmp-ecm default
08/23/11 18:14:17 v1.28.3 @ LAPPIE, Finished 1 curves using Lenstra ECM method on C64 input, B1 = 50K, B2 = gmp-ecm default
08/23/11 18:14:17 v1.28.3 @ LAPPIE, starting SIQS on c64: 7205584074334421341608887546232682732704425079530517019873416769
08/23/11 18:14:17 v1.28.3 @ LAPPIE, random seeds: 1669383867, 1919269128

bsquared · 2011-08-23, 16:47

I'm seeing it crash as well for 64 bit Windows. 32bit windows and 64 bit linux both appear to work, as does version 1.27 in 64 bit windows. That's fairly specific, so hopefully it will be quick to track down.

thanks.

Andi_HB · 2011-08-23, 17:38

Same crash with this numbers http://www.factordb.com/index.php?sh...00000452289546
http://www.factordb.com/index.php?sh...00000452296164

yoyo · 2011-08-23, 19:06

Quote:

Originally Posted by Andi_HB

I have just got an error while i am running the script from Yoyo to do some work for the DB.

Oh, since somebody is using the script, I posted my updated version in the factordb thread.

I hope you didn't run it in parallel in the same folder. This crashes yafu in my environment during the siqs phase, because both try to write the siqs.dat file.

yoyo

bsquared · 2011-08-24, 03:38

I just posted 1.28.4 to sourceforge. Bugfix version, primarily to fix the problems observed by Andi_HB and yoyo.

The siqs crashes were caused by the siqs speed upgrade of 1.28.3, so I just reverted it. It was not that much of a speed upgrade, so I doubt anyone will notice. Crashes, on the other hand, everyone notices

.

LaurV · 2011-08-24, 03:55

anyone else, except me, having trouble in resuming a nfs job with yafu in sieving phase? Yesterday before starting "the big job" on sieving that c185 fibo number, I decided to do a little bit of "rehearsal".

Fivemack seriously scared me about that 1500 hours

, so I don't want to be in a situation to resume some work after two or three days/weeks and be in impossibility to do it.

So, I started "yafu nfs(...) -siever 14 -r -threads 2 -R" on the poly I got from fivemack, let it run for half hour, hard shutdown (intentionally plugged out the computer from the socket). Hard times require hard actions

Starting again, same command, it says (in factor.log):

nfs: commencing NFS restart
nfs: previous data file found - commencing search for last special-q
nfs: parsing special-q
nfs: found xxxx relations, continuing job at specialq = xxxxxxxxx
nfs: commencing msieve filtering

Of course I could see nothing on screen, so fast off it went...):

Then I saw it created a "nfs.log" file:

error: polynomial is missing or corrupt

............
I looked around, seems like some incomplete lines on "nfs.dat". Try editing, to remove garbage characters. Restart yafu, same command. Same result. Played some more with nfs.dat, same result. I end up deleting it. Yafu working again, but sieving again from poly zero.

...............
Round two:
I said, well, maybe hard restart too tough for him...

Tried again all the procedure, but this time using ctrl+c. The "aborting..." message shows its chest on screen for a while. I said, ok, it is saving the status.

Started again. Same command line. Same result. Same log files.

Is it only me? What I am doing wrong?
I mean, before starting 1500 hours of a job, I want to make sure nothing can go wrong. Otherwise, red light for this job for a while...

bsquared · 2011-08-24, 04:40

If you have a hard restart, you'll have relations sitting in one or more rels*.dat files that will need to be manually appended to nfs.dat. If you cntl^c, yafu should do this for you (you'll know if the files disappear).

As for the corrupt poly error, open nfs.fb and remove the line containing SKEW: ###. That is something I'll need to fix, but removing the line should enable the restart to work.

Thanks for the test cases.

LaurV · 2011-08-24, 05:40

Well, it seems manageable now, after deleting the skew line.

The rels*.dat file were presented before and disappeared for the "ctrl+C" case as you said, but I did not think to mention it in my previous post. After appending them to nfs.dat with "copy nfs.dat+rels0.dat+rels1.dat nfs.dat" I saw it says "found yyyyy relations, continuing job at specialq...", where yyyy was bigger then the xxxx in previous post (it seems it found out some more relations, most probably coming from rels*.dat files).

Now just one more question: the "total yield" line is always starting from 0, is that normal? I expected to resume the counting from where it was left before crashing (pressing ctrl+c), or from a value a bit smaller (in case some relations were lost in the crashing process).

Thank a lot for your support. I started to crunch it (the c185). See ya after 1500 hours :D

jrk · 2011-08-24, 07:11

Quote:

Originally Posted by LaurV

Now just one more question: the "total yield" line is always starting from 0, is that normal? I expected to resume the counting from where it was left before crashing (pressing ctrl+c), or from a value a bit smaller (in case some relations were lost in the crashing process).

The "total yield" line is printed from gnfs-lasieve4, not Yafu. It only counts how many relations it finds each time it is run, since it has no way of knowing how many relations are saved in other files.

LaurV · 2011-08-24, 07:29

Bingo! Thanks a lot to both of you. I just begun to understand how these things work. Keeping my fingers crossed.

schickel · 2011-08-24, 08:13

Quote:

Originally Posted by LaurV

Bingo! Thanks a lot to both of you. I just begun to understand how these things work. Keeping my fingers crossed.

Well, you are kind of diving in at the deep end. Most of us start with something a little smaller.....my first gnfs job was a c130 back in March '06 (of course, that was pretty big back then.).

2011-08-23, 17:38	#806
Andi_HB Mar 2007 Germany 410₈ Posts	Same crash with this numbers http://www.factordb.com/index.php?sh...00000452289546 http://www.factordb.com/index.php?sh...00000452296164 Last fiddled with by Andi_HB on 2011-08-23 at 17:49

2011-08-24, 03:38	#808
bsquared "Ben" Feb 2007 2×3×587 Posts	1.28.4 available I just posted 1.28.4 to sourceforge. Bugfix version, primarily to fix the problems observed by Andi_HB and yoyo. The siqs crashes were caused by the siqs speed upgrade of 1.28.3, so I just reverted it. It was not that much of a speed upgrade, so I doubt anyone will notice. Crashes, on the other hand, everyone notices .

2011-08-24, 03:55	#809
LaurV Romulan Interpreter Jun 2011 Thailand 2⁶·151 Posts	anyone else, except me, having trouble in resuming a nfs job with yafu in sieving phase? Yesterday before starting "the big job" on sieving that c185 fibo number, I decided to do a little bit of "rehearsal". Fivemack seriously scared me about that 1500 hours , so I don't want to be in a situation to resume some work after two or three days/weeks and be in impossibility to do it. So, I started "yafu nfs(...) -siever 14 -r -threads 2 -R" on the poly I got from fivemack, let it run for half hour, hard shutdown (intentionally plugged out the computer from the socket). Hard times require hard actions Starting again, same command, it says (in factor.log): nfs: commencing NFS restart nfs: previous data file found - commencing search for last special-q nfs: parsing special-q nfs: found xxxx relations, continuing job at specialq = xxxxxxxxx nfs: commencing msieve filtering Of course I could see nothing on screen, so fast off it went...): Then I saw it created a "nfs.log" file: error: polynomial is missing or corrupt ............ I looked around, seems like some incomplete lines on "nfs.dat". Try editing, to remove garbage characters. Restart yafu, same command. Same result. Played some more with nfs.dat, same result. I end up deleting it. Yafu working again, but sieving again from poly zero. ............... Round two: I said, well, maybe hard restart too tough for him... Tried again all the procedure, but this time using ctrl+c. The "aborting..." message shows its chest on screen for a while. I said, ok, it is saving the status. Started again. Same command line. Same result. Same log files. Is it only me? What I am doing wrong? I mean, before starting 1500 hours of a job, I want to make sure nothing can go wrong. Otherwise, red light for this job for a while... Last fiddled with by LaurV on 2011-08-24 at 04:03

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Running YAFU via Aliqueit doesn't find yafu.ini	EdH	YAFU	8	2018-03-14 17:22
YAFU-1.34	bsquared	YAFU	119	2015-11-05 16:24
Yafu bug.	storflyt32	YAFU	2	2015-06-29 05:19
yafu-1.33	bsquared	YAFU	12	2012-11-08 04:12
yafu-1.32.1	bsquared	YAFU	21	2012-09-04 19:44

2011-08-23, 16:47	#805
bsquared "Ben" Feb 2007 3522₁₀ Posts	I'm seeing it crash as well for 64 bit Windows. 32bit windows and 64 bit linux both appear to work, as does version 1.27 in 64 bit windows. That's fairly specific, so hopefully it will be quick to track down. thanks.

2011-08-24, 04:40	#810
bsquared "Ben" Feb 2007 2·3·587 Posts	If you have a hard restart, you'll have relations sitting in one or more rels*.dat files that will need to be manually appended to nfs.dat. If you cntl^c, yafu should do this for you (you'll know if the files disappear). As for the corrupt poly error, open nfs.fb and remove the line containing SKEW: ###. That is something I'll need to fix, but removing the line should enable the restart to work. Thanks for the test cases.

2011-08-24, 05:40	#811
LaurV Romulan Interpreter Jun 2011 Thailand 2⁶·151 Posts	Well, it seems manageable now, after deleting the skew line. The rels.dat file were presented before and disappeared for the "ctrl+C" case as you said, but I did not think to mention it in my previous post. After appending them to nfs.dat with "copy nfs.dat+rels0.dat+rels1.dat nfs.dat" I saw it says "found yyyyy relations, continuing job at specialq...", where yyyy was bigger then the xxxx in previous post (it seems it found out some more relations, most probably coming from rels.dat files). Now just one more question: the "total yield" line is always starting from 0, is that normal? I expected to resume the counting from where it was left before crashing (pressing ctrl+c), or from a value a bit smaller (in case some relations were lost in the crashing process). Thank a lot for your support. I started to crunch it (the c185). See ya after 1500 hours :D

2011-08-24, 07:29	#813
LaurV Romulan Interpreter Jun 2011 Thailand 2⁶·151 Posts	Bingo! Thanks a lot to both of you. I just begun to understand how these things work. Keeping my fingers crossed.