Let's do another nice big GNFS job! - Page 6

fivemack · 2011-01-02, 21:37

68-70 is a 2M not a 200k range, but I'll stick 20 threads on it from tomorrow and should be done within a week; so let's be tidy.

fivemack · 2011-01-03, 16:28

I think I'll still be done within a week, but I would have finished quicker had I started off sieving 68-70 rather than 78-80. Well, only ten CPU-days wasted

Andi47 · 2011-01-04, 18:40

reserving 190M-190.1M for testing the windows binary of SVN-400.

Andi47 · 2011-01-06, 07:50

Quote:

Originally Posted by Jeff Gilchrist

Ok now that I'm back from my Christmas holidays, I built SVN 400, but haven't had time to test it, you can find them at the usual spot: http://gilchrist.ca/jeff/factoring/

Please let me know if it helps.

I just tested with your windows binary (32 bit) of svn-400:

gnfs-lasieve4I16e -M 1 -a 2_956+.poly -o test11.out -f 43505239 -c 100

still crashes. :(

edit: the 64 bit binary crashes too on this input.

fivemack · 2011-01-10, 09:18

180-190 arrived this morning. I'll start the linear algebra in about 36 hours (the machine is busy on a largish aliquot-sequence GNFS job), and then I'll know how long it will take.

Andi47 · 2011-01-10, 19:39

Quote:

Originally Posted by fivemack

180-190 arrived this morning. I'll start the linear algebra in about 36 hours (the machine is busy on a largish aliquot-sequence GNFS job), and then I'll know how long it will take.

You will get a last little pancake (190M-190.1M) in ~10-12 hours.

Andi47 · 2011-01-11, 05:50

Quote:

Originally Posted by Andi47

You will get a last little pancake (190M-190.1M) in ~10-12 hours.

done, 305698 relations: http://www.sendspace.com/file/ei4i2m

fivemack · 2011-01-12, 11:59

Code:

Tue Jan 11 20:41:24 2011  found 143371497 hash collisions in 538950087 relations
Tue Jan 11 20:42:04 2011  added 122178 free relations
Tue Jan 11 20:42:04 2011  commencing duplicate removal, pass 2
Tue Jan 11 21:25:50 2011  found 149500698 duplicates and 389571567 unique relations
...

Wed Jan 12 03:13:04 2011  commencing in-memory singleton removal
Wed Jan 12 03:13:06 2011  begin with 56497017 relations and 35707251 unique ideals
Wed Jan 12 03:13:18 2011  reduce to 56444972 relations and 33964109 ideals in 5 passes
Wed Jan 12 03:13:18 2011  max relations containing the same ideal: 15
Wed Jan 12 03:13:36 2011  relations with 0 large ideals: 4774664
Wed Jan 12 03:13:36 2011  relations with 1 large ideals: 16574464
Wed Jan 12 03:13:36 2011  relations with 2 large ideals: 20220231
Wed Jan 12 03:13:36 2011  relations with 3 large ideals: 11431589
Wed Jan 12 03:13:36 2011  relations with 4 large ideals: 3076722
Wed Jan 12 03:13:36 2011  relations with 5 large ideals: 357977
Wed Jan 12 03:13:36 2011  relations with 6 large ideals: 9300
Wed Jan 12 03:13:36 2011  relations with 7+ large ideals: 25
Wed Jan 12 03:13:36 2011  commencing 2-way merge
Wed Jan 12 03:13:57 2011  reduce to 42452777 relation sets and 19971914 unique ideals
Wed Jan 12 03:13:57 2011  commencing full merge
Wed Jan 12 03:17:09 2011  memory use: 1846.2 MB
Wed Jan 12 03:17:12 2011  found 22358833 cycles, need 19498328
Wed Jan 12 03:17:20 2011  weight of 19498328 cycles is about 1576453264 (80.85/cycle)
Wed Jan 12 03:17:20 2011  distribution of cycle lengths:
Wed Jan 12 03:17:20 2011  1 relations: 4774690
Wed Jan 12 03:17:20 2011  2 relations: 2296096
Wed Jan 12 03:17:20 2011  3 relations: 1782301
Wed Jan 12 03:17:20 2011  4 relations: 1490219
Wed Jan 12 03:17:20 2011  5 relations: 1356786
Wed Jan 12 03:17:20 2011  6 relations: 1229695
Wed Jan 12 03:17:20 2011  7 relations: 1138510
Wed Jan 12 03:17:20 2011  8 relations: 1035215
Wed Jan 12 03:17:20 2011  9 relations: 943658
Wed Jan 12 03:17:20 2011  10+ relations: 3451158
Wed Jan 12 03:17:20 2011  heaviest cycle: 17 relations
Wed Jan 12 03:17:43 2011  commencing cycle optimization
Wed Jan 12 03:18:09 2011  start with 100797206 relations
Wed Jan 12 03:20:35 2011  pruned 9176816 relations
Wed Jan 12 03:20:35 2011  memory use: 2910.8 MB
Wed Jan 12 03:20:35 2011  distribution of cycle lengths:
Wed Jan 12 03:20:35 2011  1 relations: 4774690
Wed Jan 12 03:20:35 2011  2 relations: 2401239
Wed Jan 12 03:20:35 2011  3 relations: 1934439
Wed Jan 12 03:20:35 2011  4 relations: 1667350
Wed Jan 12 03:20:35 2011  5 relations: 1563128
Wed Jan 12 03:20:35 2011  6 relations: 1428863
Wed Jan 12 03:20:35 2011  7 relations: 1315920
Wed Jan 12 03:20:35 2011  8 relations: 1150431
Wed Jan 12 03:20:35 2011  9 relations: 983890
Wed Jan 12 03:20:35 2011  10+ relations: 2278378
Wed Jan 12 03:20:35 2011  heaviest cycle: 17 relations
Wed Jan 12 03:21:10 2011  RelProcTime: 31060
...

Wed Jan 12 04:44:19 2011  matrix is 19222030 x 19222274 (5858.3 MB) with weight 1504592773 (78.27/col)
Wed Jan 12 04:44:19 2011  sparse part has weight 1343506186 (69.89/col)
Wed Jan 12 04:44:19 2011  using block size 65536 for processor cache size 8192 kB
Wed Jan 12 04:45:45 2011  commencing Lanczos iteration (4 threads)
Wed Jan 12 04:45:45 2011  memory use: 5213.1 MB
Wed Jan 12 04:49:07 2011  linear algebra at 0.0%, ETA 683h14m
...
11743 nfsslave  20   0 7000m 6.8g 1000 R  396 57.8   2136:42 msieve-td130  
linear algebra completed 205333 of 19222274 dimensions (1.1%, ETA 668h57m)

I don't understand how the weight has ended up so low, since I think I built this msieve binary with TARGET_DENSITY=130. Anyway, factors should be here by Valentine's Day, and I can start thinking of new GNFS jobs in comparison to which this one is the least of bagatelles.

Andi47 · 2011-01-12, 12:05

@fivemack: have you got (and included) my relations from 190-190.1M?

Yes

jasonp · 2011-01-12, 13:01

Quote:

Originally Posted by fivemack

I don't understand how the weight has ended up so low, since I think I built this msieve binary with TARGET_DENSITY=130

Maybe it's time to rethink the merging rules; merging stops when either the matrix is dense enough or no more columns can be deleted. Your matrix size is quite close to the number of ideals remaining after the 2-way merge, so this is a case where there was so much oversieving that the merge went as far as it could.

fivemack · 2011-02-08, 19:46

I regret there will be a substantial delay before the factorisation of this number is completed: the Lanczos step produced only trivial kernel elements. I'll try again with rather fewer input relations and a rather lower bound, but that will take at least two months.

2011-01-12, 12:05	#64
Andi47 Oct 2004 Austria 2·17·73 Posts	@fivemack: have you got (and included) my relations from 190-190.1M? Yes Last fiddled with by fivemack on 2011-01-12 at 13:19 Reason: answered question

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Nice progress!	schickel	FactorDB	29	2012-07-18 17:03
Nice pic	Dubslow	Forum Feedback	0	2012-05-02 02:13
The Hello - I am - Nice to meet you thread....	Prime Monster	Lounge	23	2012-02-11 11:08
Very nice strategical puzzle	Raman	Puzzles	2	2009-11-01 10:23
Nice link...	Xyzzy	Lounge	4	2003-06-28 13:37

2011-01-02, 21:37	#56
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 1100100010111₂ Posts	68-70 is a 2M not a 200k range, but I'll stick 20 threads on it from tomorrow and should be done within a week; so let's be tidy.

2011-01-03, 16:28	#57
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 3×2,141 Posts	I think I'll still be done within a week, but I would have finished quicker had I started off sieving 68-70 rather than 78-80. Well, only ten CPU-days wasted

2011-01-04, 18:40	#58
Andi47 Oct 2004 Austria 2·17·73 Posts	reserving 190M-190.1M for testing the windows binary of SVN-400.

2011-01-10, 09:18	#60
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 1917₁₆ Posts	180-190 arrived this morning. I'll start the linear algebra in about 36 hours (the machine is busy on a largish aliquot-sequence GNFS job), and then I'll know how long it will take.

2011-02-08, 19:46	#66
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 3·2,141 Posts	I regret there will be a substantial delay before the factorisation of this number is completed: the Lanczos step produced only trivial kernel elements. I'll try again with rather fewer input relations and a rather lower bound, but that will take at least two months.