Completed 100 curves at B1 = 110M and B2 = 680270182898, for M1061.

I did stage 1, xyzzy did stage 2.

Thomas
:coffee:

Completed 400 curves: B1=11e7, B2=52e9, Dickson(12) polynomial (multiplier 0.61)

I don't want to intrude on the posts or discussion here, but I have a question. How many of these "curves" do you have to perform before you are finished searching a number or a range?

[QUOTE=Jwb52z]I don't want to intrude on the posts or discussion here, but I have a question. How many of these "curves" do you have to perform before you are finished searching a number or a range?[/QUOTE]
Read the beginning of this thread. :)

[QUOTE=cheesehead]Read the beginning of this thread. :)[/QUOTE]

Not being a math person, all that table at the beginning of the thread tells me is that there are levels to the factoring and it can keep going up. That is why I asked the question. When is the number completely factored and finished forever?

[QUOTE=Jwb52z]Not being a math person, all that table at the beginning of the thread tells me is that there are levels to the factoring and it can keep going up. That is why I asked the question. When is the number completely factored and finished forever?[/QUOTE]

A number is completely factored if we find all it's prime factors.

There are different methods to factor a number. One of them is ECM, which can found 50-55 digit primefactors with current bounds. You just run several thousands relatively short curves and hope to get a lucky one that finds a factor (if exists around that size).

Another method to factor the number is NFS. But NFS on numbers this size takes several GHz years of sieving and requires a very powerful machine with lots of memory to do the post processing. M1061 is well above the current NFS record

Time for the first M1061 summary of 2005!
Previous summary: 729 of 22000 curves done.

[CODE]
username: curves mult B2 Message no Line total:
xyzzy 67 1 680270182898 #170 67
xyzzy+tn 10 1 680270182898 #184 10
thomasn 100 0.35 4290000000 #187 35
xyzzy+tn 100 1 680270182898 #188 100
geoff 400 0.61 52e9 #189 244
total this period 456

grand total :
1185 of 22000 curves done.
[/CODE]

Steady progress. :smile:

Thomas
:coffee:

[QUOTE=Jwb52z]Not being a math person, all that table at the beginning of the thread tells me is that there are levels to the factoring and it can keep going up. That is why I asked the question. When is the number completely factored and finished forever?[/QUOTE]Jwb52z,

I apologize for my flippant response to your question. I didn't take the time to properly consider what might be most helpful to you.:redface:

Richard Woods

100 curves B1=110 000 000 B2=680270182898.

I did stage 1, Xyzzy did stage 2.

Thomas
:coffee:

200 curves with B1=11e7, B2=52e9, degree 12 polynomial (multiplier 0.61).

I have some more stage one residues ready but I will wait until ECM6 comes out to complete them, hopefully it will make a larger stage two bound worthwhile.

I guess we should be reporting these curves in the new Cunningham Tables forum, they have a column for 55 digit curves.

Doing stage two with gmp-ecm 6.0 instead of 5.0.3 increases the total productivity for this project by 24% on my P4, very impressive considering stage two only accounts for one third of the total computation.

The optimal B2 has increased from 52 billion to 198 billion (when using mprime for stage one). Here are some timings, stage one with B1=11e7 takes 1368s using mprime, the final column is the number of days to complete the 55 digit level on this machine using the (reduced) number of expected curves reported by gmp-ecm 6.0:
[code]
B2' Stage2 Memory Total Expected Days to find
Time Time Curves 55 digit factor
-------------------------------------------------------
62e9 327s 394MB 1695s 28835 566d
111e9 460s 529MB 1828s 25857 547d
149e9 547s 620MB 1915s 24488 543d
198e9 620s 720MB 1988s 23287 536d *
251e9 721s 778MB 2089s 22380 541d
332e9 1583s 1073MB 2951s 19531 667d
[/code]
I have 1GB in this machine so the last run resulted in some minor swapping. A default curve with B2=680e9 uses too much memory (> 1400MB?) for this machine, but would not be worthwhile anyway.