Factoring humongous Cunningham numbers

[bdodson]

Originally Posted by FactorEyes

Decent-sized ECM hit after about 6000 curves at B1=11e7.

********** Factor found in step 2: 18435808944826087315724004590959692083957933332391173084977217
Found probable prime factor of 62 digits:

...

That's 3^496+2^496. Very nice! Still not on the ecmnet page; bumps
a p60, making the 10th of the top10 a p61. Next p62-or-larger makes
the 10th a p62! Then we can start worrying which p62 gets bumped.

As I don't believe that the email was intended as a private communication
(sent to ten people), PaulZ reports

The first time we had no p4x any more in the yearly top ten was in 2003.
Thus we gained 10 digits in 7 years.

along with the note that it's still only May! -Bruce

[Xyzzy]

…and you're living under a bridge in a van, down by the river.

FTFY

[xilman]

Another update has just been posted. It contains 11 factorizations, 10 of which are complete. There are now 740 composites left in the tables.

Paul

[xilman]

Originally Posted by xilman

Heads up everybody.

Phil Carmody has very generously hosted an ECMnet server for me on his home network. He has been running it for several years; the server currently contains the homogeneous Cunningham numbers.

Phil has now told me that he will be moving home some time in the next few months and can not guarantee that he can host the server afterwards. Whether or not the latter is the case, there will be interruptions.

We are doing what we can to minimize the inconvenience --- this advance notice is part of that process --- but there may well be some problems.

It's possible that the server may migrate to a system under my direct control. If that happens, the IP name and address will definitely change. However, I'll do what I can to ensure a smooth changeover.

Paul

Phil has just told me that the changeover may occur as soon as this Friday.

Sorry to be quite so emphatic but the notice is now really quite short. Phil says he may be able to keep the server running on 83.143.57.194:8194 for a shortish period. While that would be nice if it's possible, please don't bank on it.

It's not clear what will be the long-term arrangements. I'm running an ECMnet server on the standard port (8194) on my home machine which has IP address 83.217.167.177 and you're welcome to connect there during the interregnum. However, it is loaded with generalized Cullen and Woodall numbers at the moment.

The homogeneous Cunningham tables will be hosted somewhere when things settle down.

Paul

[joral]

Originally Posted by xilman

Phil has just told me that the changeover may occur as soon as this Friday.

Sorry to be quite so emphatic but the notice is now really quite short. Phil says he may be able to keep the server running on 83.143.57.194:8194 for a shortish period. While that would be nice if it's possible, please don't bank on it.

It's not clear what will be the long-term arrangements. I'm running an ECMnet server on the standard port (8194) on my home machine which has IP address 83.217.167.177 and you're welcome to connect there during the interregnum. However, it is loaded with generalized Cullen and Woodall numbers at the moment.

The homogeneous Cunningham tables will be hosted somewhere when things settle down.

Paul

What is needed to run the server? I have a server out there which isn't doing much, so I could always host it (depending on how much work is involved.)

[xilman]

Originally Posted by joral

What is needed to run the server? I have a server out there which isn't doing much, so I could always host it (depending on how much work is involved.)

Thanks for the offer.

If you can let me have shell access to your server, essentially no work is involved. This is the arrangement I've had with Phil for the last few years. Otherwise it can get a bit tricky because the server has to be kept in sync with factors found by other means --- principally NFS factorizations but occasional non-ECMnet ECM runs.

Regardless of access permissions, the host needs to have available a few megabytes of disk space, a working C++ compiler and unblocked access to port 8194/TCP. The cpu usage is barely noticeable, being a second or two per day. I've very little idea about the IP data traffic other than I've never noticed it being significant.

Please feel free to contact me by email if you wish to take this further.

Phil's latest news, BTW, is that his server will be able to continue beyond tomorrow but doesn't know for how much longer.


Paul

[fatphil]

Originally Posted by xilman

Phil's latest news, BTW, is that his server will be able to continue beyond tomorrow but doesn't know for how much longer.

Greetings all. Sorry for any inconvenience this has caused.

I've got a replacement machine that literally fits in my pocket that can take over the role at a moment's notice, and which I can keep running well into mid-late june. (Same IP address, same port, only Paul's password changes ;-) )

The sooner I decommission this big box, the better, so migrating onto that replacement is good for me. (And of course, if I can get set up in Tallinn quickly, I might even be able to hand back over to myself!)

Regarding resource consumption - it's negligible. Less than 0.0001% of what SMTP uses.
(That'll teach me to have domain name 'asdf.org' ;-) )

[fatphil]

Originally Posted by xilman

unblocked access to port 8194/TCP.

If the clients are going to need to be reconfigured anyway, then they can connect to another port instead. 8194 is simply tradition.

[xilman]

Originally Posted by fatphil

(And of course, if I can get set up in Tallinn quickly, I might even be able to hand back over to myself!)

Joral has set up an account on a server for me to migrate the ECMnet server on to. It will take me a little time to get everything settled in but I hope to do that over the next few days. Another business trip starting Tuesday doesn't help in that respect.

My plan is to convert Phil's server into a slave of Joral's and verify that everything mostly works most of the time (pessimist, moi?) before telling everyone to switch servers. Assuming Phil's box continues for a couple of weeks the changeover should be relatively smooth.


Paul

[fatphil]

Originally Posted by xilman

Joral has set up an account on a server for me to migrate the ECMnet server on to. It will take me a little time to get everything settled in but I hope to do that over the next few days. Another business trip starting Tuesday doesn't help in that respect.

My plan is to convert Phil's server into a slave of Joral's and verify that everything mostly works most of the time (pessimist, moi?) before telling everyone to switch servers. Assuming Phil's box continues for a couple of weeks the changeover should be relatively smooth.
Paul

In that case, I will need to perform the migration onto my PicoITX. It looks like the ecmserver binary runs happily (though I've not seen it do any GMP maths yet, the VIA might not like the might-be-built-for-athlon library). It's late, so I'll leave this till tuesday.

Hopefully in the space of no more than a couple of minutes I can then perform the following:
(as you) stop your ECM server instance;
rsync your home directory deltas to the picoITX;
pull .194 networking down;
pull the picoITX network up as both .194 and its normal IP addresses;
(as you) start the ECM server again.
wait to see some clients connect.
mail you your new password!

[jyb]

Originally Posted by xilman

My plan is to convert Phil's server into a slave of Joral's and verify that everything mostly works most of the time (pessimist, moi?) before telling everyone to switch servers. Assuming Phil's box continues for a couple of weeks the changeover should be relatively smooth.


Paul

Thanks for the update, Paul. Can I put in a plug for using this opportunity to upgrade the server to a more recent version (I believe the latest is 2.7.3)? It would be really nice to be able to see the server's current tables via a web browser.