now on to 2^757-1

Paulie · 2003-07-31, 14:16

Quote:

Originally Posted by wblipp

2^713-1 reached the target of 43 million relations. The server is now handing out assignments for 2^757-1.

So how many relations will this project need and our expected completion time?

It's getting to the point where the sieving is being done so fast that the cluster doing the actual work is falling behind a few projects. :)

Wacky · 2003-07-31, 15:25

The early estimates are something like 24M lines, 76M relations, and 14K Work units.

Rough estimate -- Late September if we don't get significantly more participation.

The post-processing takes only about 2 weeks of cluster time. Although it may get behind temporarily when we do an "easy" factoring, it can catch up when we tackle the harder ones like 10^277- or 2^757- .

Paulie · 2003-07-31, 18:21

Cool. I love this project for two reasons: 1) running my G4/800 Mac isn't embarrasing and 2) there's something new a month or so down the pipe. :) :) :)

Thanks for letting us play in your sandbox!!! :D

Wacky · 2003-07-31, 19:46

Quote:

Originally Posted by Paulie

Thanks for letting us play in your sandbox!!! :D

We are happy to have you. Bring some friends. There is plenty of sand.

We have hundreds of miles of beach here in Texas. And, if that is too warm to suit you, I also know a serving wench in Britain who can keep you busy. ( But you won't find her in the /pub ;) )

xilman · 2003-08-01, 14:27

Quote:

Originally Posted by Paulie

It's getting to the point where the sieving is being done so fast that the cluster doing the actual work is falling behind a few projects. :)

Yes and no. The cluster is still working on 10,227- but that's because the earlier project was so much harder than the one which has just finished sieving. The easier one caught up, if you like.

In a few weeks the situation should have reversed. The post-processing for 10,227- will have finished and so will the much easier 2,713-. Sieving 2,757- will still be chugging along, unless there's a sudden and massive influx of new contributors.

Perhaps I should mention it here: the linear algebra stage for 10,227- is predicted to finish at 1pm (Cambridge time) on 15th August. That's assuming no hold-ups occur and that progress doesn't speed up but continues at the same rate as it has for the last several days. The former can't be guaranteed but the latter seems to be a reasonable assumption.

There's no intrinsic problem with this time and date. The only reason I quote it so precisely is that I'm going on vacation for two weeks and will leave work at about 5pm on the same day. I may get intermittent email and web access but I won't be able to contact the cluster while I'm away.

If the linear algebra finishes in time, I will be able to start the matrix for 2,713- and leave it running while I am out of the office. I will also be able to kick off the square root phase and, with luck, get a factorization.

If I run out of time, we will have to wait until the beginning of September for the factors and to start the next matrix.

Paul

smh · 2003-08-17, 08:57

Quote:

Originally Posted by xilman

Perhaps I should mention it here: the linear algebra stage for 10,227- is predicted to finish at 1pm (Cambridge time) on 15th August. That's assuming no hold-ups occur and that progress doesn't speed up but continues at the same rate as it has for the last several days. The former can't be guaranteed but the latter seems to be a reasonable assumption.

There's no intrinsic problem with this time and date. The only reason I quote it so precisely is that I'm going on vacation for two weeks and will leave work at about 5pm on the same day. I may get intermittent email and web access but I won't be able to contact the cluster while I'm away.

No news on the website, so i guess you didn't make it?

Wacky · 2003-08-17, 09:12

No, the last prediction that I saw was for about 9PM. Unfortunately, Paul left a little after 5. I do know that it finished because the processors show a noticable increase in sieving speed.

Except for the delay in starting the square root phase and getting the answer, it's not a problem. After Paul returns, he still has enough time to catch up before we finish sieving M(757).

xilman · 2003-09-01, 14:29

Quote:

Originally Posted by Wackerbarth

No, the last prediction that I saw was for about 9PM. Unfortunately, Paul left a little after 5. I do know that it finished because the processors show a noticable increase in sieving speed.

Except for the delay in starting the square root phase and getting the answer, it's not a problem. After Paul returns, he still has enough time to catch up before we finish sieving M(757).

We caught up. Subscribers to nfsnet-l will have seen the factors of both 10,227- and 2,713- by now.

The more formal announcements will be made in the very near future, probably when Don has recovered from Labor Day --- which is something I don't get on this side of the Atlantic. I get May Day instead.

Paul

2003-07-31, 15:25	#2
Wacky Jun 2003 The Texas Hill Country 441₁₆ Posts	The early estimates are something like 24M lines, 76M relations, and 14K Work units. Rough estimate -- Late September if we don't get significantly more participation. The post-processing takes only about 2 weeks of cluster time. Although it may get behind temporarily when we do an "easy" factoring, it can catch up when we tackle the harder ones like 10^277- or 2^757- .

2003-07-31, 18:21	#3
Paulie Aug 2002 DF₁₆ Posts	Cool. I love this project for two reasons: 1) running my G4/800 Mac isn't embarrasing and 2) there's something new a month or so down the pipe. :) :) :) Thanks for letting us play in your sandbox!!! :D

2003-08-17, 09:12	#7
Wacky Jun 2003 The Texas Hill Country 1089₁₀ Posts	No, the last prediction that I saw was for about 9PM. Unfortunately, Paul left a little after 5. I do know that it finished because the processors show a noticable increase in sieving speed. Except for the delay in starting the square root phase and getting the answer, it's not a problem. After Paul returns, he still has enough time to catch up before we finish sieving M(757).