Thread: Machines
View Single Post
Old 2004-06-26, 09:41   #2
xilman's Avatar
May 2003
Down not across

11,027 Posts

Originally Posted by Bob Silverman

225 machines doing sieving for NFSNET is actually a very small number
when compared with the (approximately) 200,000 associated with GIMPS.

Let's start a campaign to get more help. Unlike the linear algebra phase,
sieving can use all the help it can get....

As you note, sieiving is one of the easier portions of the entire NFS process. It may require the most cpu cycles but it's relatively easy to perform. Indeed, once the basic number-theoretical code is written and working relatively reliably, the mathematical process as a whole is relatively easy to perform.

Some other phases are much more difficult. The linear algebra is sometimes a bottleneck, but the real reason at present why NFSNET can not expand its operations greatly is lack of human resources: for code development, project administration, web site maintenance, documentation writing and doubtless more that I've not mentioned.

At the moment Richard Wackerbarth is shouldering a large fraction of the load. Too large a fraction. I've been trying to help where I can but a combination of legal constraints, developments at work and personal reasons outside work means that I've not been able to do as much as I would like or as much as I should.

Do you want to help NFSNET in one or more of these areas?

Does any other reader of this forum want to assist?

I would very much like to see thousands of machine sieving, but without back-office support it's not going to be possible.

xilman is offline