10^263-1 sieving

[Xyzzy]

Xyzzy, smh: keep running your ranges until they are done.

We are about 2 or 3 days from being done.

[smh]

2354284 relations are on it's way to Oslo as i write this.

[fivemack]

Excellent. I will start baking the relations this evening; I think with 15M-100M we were very close to a matrix, so there should be enough now for a reasonably-sized one.

[fivemack]

ETA for the matrix

Sat Feb 28 08:20:51 2009  matrix is 15785588 x 15785836 (4555.7 MB) with weight 1111879653 (70.44/col)
Sat Feb 28 08:20:51 2009  sparse part has weight 1036401865 (65.65/col)

is April 2nd; maybe this could have done with a bit more over-sieving, but it takes quite a lot of time for a request for over-sieving to get answered, so I doubt I'd have collected in a week enough more relations to get the linalg time down from 33 days to 26.

[Wacky]

I doubt I'd have collected in a week enough more relations to get the linalg time down from 33 days to 26.

Tom,

Is the intent to produce a particular result ASAP? Or, as I view the NFSNet effort, to maximize the results over time. In the latter case, the ability to start, stop, and restart sieving effort (and providing work for the "next" project during the down time), is critical to the efficient use of resources. The largest problem that I currently encounter is that too few of the resources are participating in an automated manner. This causes me to expend far too much of the really critical resource (personal time) to incorporate potential automated resources.

Richard

[fivemack]

40199712048008930571428131875902950777477607917 * 44098347924570538726555032711112035074551402751062500493727027459523924634267 * 6267758127577332259078470898197707371104920617239537855165700322603688636346683832426409007658616045014902930862175109617951881840830949049
%5 = 11111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111

Some highlights (full log attached):

Fri Feb 27 19:57:50 2009  commencing relation filtering
...
Fri Feb 27 21:20:54 2009  found 113044024 duplicates and 171805386 unique relations
...
Sat Feb 28 03:57:26 2009  weight of 15859152 cycles is about 1110269217 (70.01/cycle)
...
Sat Feb 28 08:20:51 2009  matrix is 15785588 x 15785836 (4555.7 MB) with weight 1111879653 (70.44/col)
Sat Feb 28 08:20:51 2009  sparse part has weight 1036401865 (65.65/col)
Sat Feb 28 08:20:51 2009  matrix includes 64 packed rows
Sat Feb 28 08:20:52 2009  using block size 65536 for processor cache size 2048 kB
Sat Feb 28 08:22:03 2009  commencing Lanczos iteration (4 threads)
Sat Feb 28 08:22:03 2009  memory use: 4861.9 MB
...
Thu Apr  2 08:13:41 2009  lanczos halted after 249630 iterations (dim = 15785585)
Thu Apr  2 08:14:30 2009  recovered 33 nontrivial dependencies
...
Thu Apr  2 21:18:10 2009  reading relations for dependency 5
Thu Apr  2 21:18:22 2009  read 7895094 cycles
Thu Apr  2 21:19:59 2009  cycles contain 27711405 unique relations
Thu Apr  2 21:29:50 2009  read 27711405 relations
Thu Apr  2 21:38:34 2009  multiplying 22852064 relations
Thu Apr  2 22:38:16 2009  multiply complete, coefficients have about 598.57 million bits
Thu Apr  2 22:38:32 2009  initial square root is modulo 234499
Fri Apr  3 00:22:06 2009  prp77 factor: 44098347924570538726555032711112035074551402751062500493727027459523924634267
Fri Apr  3 00:22:06 2009  prp139 factor: 626775812757733225907847089819770737110492061723953785516570032260368863634668383242640900765861604501490293086217510\
9617951881840830949049

Quite a large matrix - a smidgen smaller than the C180 GNFS. 820 hours on four cores of 2.5GHz Phenom to do the linalg steps. The duplicate rate was quite high by the end of sieving; I guess total sieving effort was 46,000 C2/2400-hours = 165Msec (IE fifteen times longer than the linalg).

Wagstaff and Kamada have been informed.

[Batalov]

Congratulations! 31-bit projects rule.

Page 110 gets pretty full. (I wonder if it will instead open the page 111.)

[R.D. Silverman]

Originally Posted by fivemack

[code]
40199712048008930571428131875902950777477607917 * 44098347924570538726555032711112035074551402751062500493727027459523924634267 * 6267758127577332259078470898197707371104920617239537855165700322603688636346683832426409007658616045014902930862175109617951881840830949049

Wagstaff and Kamada have been informed.

Nicely done. Now if we could only tackle the 1st few holes in the
base 11 and 12 tables........ These have languished for some time.

[xilman]

Originally Posted by R.D. Silverman

Nicely done. Now if we could only tackle the 1st few holes in the
base 11 and 12 tables........ These have languished for some time.

You called?

I may be able to help out there. The current work on 3,508+ has reached the filtering stage. Depending on the matrix, factors may be ready before the end of the month.

In the mean time, my sievers are idle (actually, they are running ECM on the Cullen and Woodall tables) and could do with a new project. I'd been thinking of a 800-bit C or W SNFS but may change my mind.


Paul

[R.D. Silverman]

Originally Posted by xilman

You called?

I may be able to help out there. The current work on 3,508+ has reached the filtering stage. Depending on the matrix, factors may be ready before the end of the month.

In the mean time, my sievers are idle (actually, they are running ECM on the Cullen and Woodall tables) and could do with a new project. I'd been thinking of a 800-bit C or W SNFS but may change my mind.


Paul

My CPU resources max out at around 780 bits (unless I want to spend 6
months on one number)........