Where the heck are the stats?

[jasong]

I tried the obvious thing of Googling "nfsnet stats" but the first two pages didn't get me what I wanted, so I gave up.

If there are no official stats, perhaps someone could inform me how many lines we're going for with 6^283-1, and about how much progress is made each day.

[Wacky]

The stats have been down for some time. Recently someone indicated a willingness to try to get them up again. He was rewriting some of the code to work on Linux rather than Windoze. I do know that we are sending him the raw data each day. I hope to see something soon.

As for 6,283-, I project that we will sieve 20M lines and have done 1/3 of them already. Based on the current rate, we should finish around July 5, but that is likely to slip a few days because of wayward participants.

[R.D. Silverman]

Originally Posted by Wacky

The stats have been down for some time. Recently someone indicated a willingness to try to get them up again. He was rewriting some of the code to work on Linux rather than Windoze. I do know that we are sending him the raw data each day. I hope to see something soon.

As for 6,283-, I project that we will sieve 20M lines and have done 1/3 of them already. Based on the current rate, we should finish around July 5, but that is likely to slip a few days because of wayward participants.

Although daily, real-time stats would be "nice", they really are not necessary.
A simple text posting once a week would suffice for me (and I suspect most
people). For example, the last announcement was about 2,772+ being
70% done. i.e. well out of date. It isn't even mentioned that 6,283- is active.

[bdodson]

Originally Posted by R.D. Silverman

Although daily, real-time stats would be "nice", they really are not necessary.
A simple text posting once a week would suffice for me (and I suspect most
people). For example, the last announcement was about 2,772+ being
70% done. i.e. well out of date. It isn't even mentioned that 6,283- is active.

Me too. For example, the person with access to the daily stats
(as discussed with frm_Kentucky elsewhere) might post a final count
from 2,772+ either with or without the last of the lattice-siever data.
I'd be interested to see whether I hit 50% on this one. Only downside
might be that we'll shortly have inquiries as to when to expect factors
(more than a week (fairly certain), less than a month (way less, we hope)).

On 6,283- (which WAS mentioned in my most recent post on the 772-has-started
thread, 1st paragraph; in a 2nd reply/follow_up to Bob's post on the open 768-bit
candidates), some of the nfsnet software seems to have been written assuming
that daily stats would be available. A positive point from Tom's efforts,
if one scrolls all the way down on the main nfsnet page (linked on the
forum listing for this discussion), one finds a further link to "more news", at

http://www.nfsnet.org/status.html

which now includes copies of the original stats final stats for factorizations
up through April 2005. If one clicks on one of the projects, it's easy to see
why this was a great resource (when it ran well). For example, clicking
further on "users" and on "computers" shows each nfsnet guid for each
machine; and when I started (relatively recently) the info was updated daily.
Not that it's supposed to matter to us fanatics, but it's a very nice
instance of a program that produces a graphical version of the info.
Some sort (minimial even) of a recovery of the software from "bit rot"
would seem to be a Very good thing. -Bruce

[xilman]

Originally Posted by R.D. Silverman

Although daily, real-time stats would be "nice", they really are not necessary.
A simple text posting once a week would suffice for me (and I suspect most
people). For example, the last announcement was about 2,772+ being
70% done. i.e. well out of date. It isn't even mentioned that 6,283- is active.

Very well.

The linear algebra for 2,772+ started a few hours ago and is predicted to take just under 19 days to complete ---- that's cpu time and not elapsed time, though as the program is currently averaging 99% cpu the two may not may be too different.

The matrix is 5.6M square and has close to 80 bits per row and per column. The Lanczos run is using 1989MB active virtual memory.


Paul