2^772+1 has started

[Wacky]

Originally Posted by R.D. Silverman

What's the current status?

Sieving on 6,283- has now reached 67%. At the current production rate, we should finish about July 9.

However, note that a significant number of the sievers are "without supervision" for much of that time. As a result, we may lose their service at any time and not have it restored for some days.

As a result, I would suspect that the projected date is "optimistic".

[fivemack]

Out of curiosity, what's the weight of the matrix (after removing the dense rows) for 2+772? Presumably less than 2^29, since it fits in xilman's machine which IIRC has 2GB.

[xilman]

Originally Posted by fivemack

Out of curiosity, what's the weight of the matrix (after removing the dense rows) for 2+772? Presumably less than 2^29, since it fits in xilman's machine which IIRC has 2GB.

443M set bits after removing the heaviest rows.

Only 200K vectors to go now (out of 5.54M) so should be finished in a day or so.

My machine has 2.5G RAM and the extra 0.5G allows the machine to run other things, such as an operating system, without paging itself to a standstill when running a 2G AVM process. The Lanczos process is taking 1989M AVM and yet the machine is still very responsive as long as long as I don't fire up anything that requires a lot of memory.


Paul

[PBMcL]

I'm not sure why Paul hasn't posted it here, but Wagstaff is showing the result for 2^772+1 at

http://homes.cerias.purdue.edu/~ssw/cun/page105

c215 = p107 * p108, a very nice split.

[xilman]

Originally Posted by PBMcL

I'm not sure why Paul hasn't posted it here, but Wagstaff is showing the result for 2^772+1 at

http://homes.cerias.purdue.edu/~ssw/cun/page105

c215 = p107 * p108, a very nice split.

Because I'd asked Wacky to do so.


Paul

[bdodson]

Originally Posted by xilman

Because I'd asked Wacky to do so.
Paul

In fairness, there was a response to Paul's request for
"appropriate announcements to the NFSNET people". Factoring
announcements from all of the other factoring groups include,
not just the matrix and square-root information (the factors!), but
some info on the sieving. On the last several NFSNET factorizations
some two thirds of the relations were found on machines at the
Univ of Gent and at Lehigh. The Gent machines were busy elsewhere
for most of the sieving for 2,772+, so that (perhaps) half of the
relations were found at Lehigh.

I've been suggesting that the official record in the text acompanying
Sam's pages ought to reflect these institutional contributions. I'd
actually be somewhat happier if the first report anyone outside of
NFSNET sees of our factors included everyone supplying 3% or more
of the sieving, for motivation of our contributors, if nothing else. So
the delay in the report may be more my fault than Richard's.

On news of more general interest, sieving for 6, 283- seems near to
being done (early, due to the return of the Gent machines). In particular,
today's project files have started the sieving on the next NFSNET number,
5,317-. As has been our recent practice, polynomials and sieving
parameters are due to Paul, while Richard does the part of filtering
that prunes the data for file transfer. Matrix building (with final merges
in the filtering) is usually done by the person running the matrix; so Paul
for 2,772-.
-bdodson (very glad to have these c10x prime factors out of the
ecmnet/cunningham queue!)