[QUOTE=philmoore]But isn't just one square-root computation of the same overall complexity as a complete LL test?[/QUOTE]

Correct. Faking the save file will take as much cpu time as 5000 LL tests of similar exponent size. What is worse, you need the complete factorisation of Mp in order to compute square roots (mod Mp).

Alex

Cross your fingers !
 

The check is running well.

However, please all of you cross your fingers in order that:
- no one reboots the machine,
- we do not fall into a hardware bug or into a Linux bug,
- we do not fall into a GLucas hang problem I've seen some times in the past.

Before launching the check, I've run GLucas self-tests, which ran perfectly !!
Now the POSIX Threads NPTL system is much more robust than previously.
The machine has been used by colleagues for other long and complex tasks.

So I am really optimistic !

Tony

[QUOTE=T.Rex]The check is running well.

However, please all of you cross your fingers in order that:
- no one reboots the machine,
- we do not fall into a hardware bug or into a Linux bug,
- we do not fall into a GLucas hang problem I've seen some times in the past.

Before launching the check, I've run GLucas self-tests, which ran perfectly !!
Now the POSIX Threads NPTL system is much more robust than previously.
The machine has been used by colleagues for other long and complex tasks.

So I am really optimistic !

Tony[/QUOTE]

we cross, we cross . :smile:
Does GLucas save periodically its results ? If somebody reboots the machine (or a sudden power down comes) can it recover from last savefile ? I hope yes ...

For those who worry about extra verification, I started an Mlucas run on a HP Visualize C300 with a PA-RISC processor which will take about 250 days to complete. That is if I guessed the right exponent :surprised

[QUOTE=pacionet]we cross, we cross . :smile:
Does GLucas save periodically its results ? If somebody reboots the machine (or a sudden power down comes) can it recover from last savefile ? I hope yes ...[/QUOTE]Yes, GLucas saves its results every 50000 iterations (into 2 flip-flop files). Also, I've configured GLucas to save an "interim" file every 1 million iterations. And I've put in place a "cron" that copies all the files from the test machine to my PC at work every half an hour.
So, if the machine crashes or if someone reinstalls the machine, there is no problem to restart the check on another machine (if I can find one other and if I am in town at that time ... If not, then Jeff will provide the result of his check first !).
Tony

Mathworld has a news item on this:

[url]http://mathworld.wolfram.com/news/2005-12-19/mersenne-43/[/url]

Strangely enough, the article lists M13466917 as the 39th Mersenne prime without any question mark, even though it admits later on that not all exponents less than 13466917 have been double-checked.

Also, here's an interesting quote:

"While the exact exponent of the new find has not yet been made public, GIMPS organizer George Woltman reported that the new candidate has fewer than 10 million digits (a holy grail for prime searchers), meaning that the new candidate has exponent n somewhere between 24,036,584 and 33,219,253. Because Woltman conspicuously did not state that the current candidate would be the largest known prime, it is likely that the 43rd known Mersenne prime is smallerthan the 42nd. Woltman is currently attempting to reproduce the find from the user's save file, thus eliminating any chance of the report being erroneous. "

[QUOTE]Does GLucas save periodically its results ? If somebody reboots the machine (or a sudden power down comes) can it recover from last savefile ? I hope yes ...[/QUOTE]

In addition to the save and interims files, I put some effort doing the files universaly usable. So the itanium intermediate result files can be used by an alpha, a pentimu4 or most of the current popular processors. :rolleyes:

You can see the format of save files here:

[url]http://www.oxixares.com/glucas/doc/Save-file-format.html#Save%20file%20format[/url]

Guillermo

Since George and Guillermo seem not being around, I announce:
Verification is two-third complete.

Do you think in will be done before the end of the 24th?

Alex

[quote]Ha, you should call that OS hybrid "RedBull Advanced Server." Of course you'd immediately get sued for trademark infringement[/quote]
Or maybe not.

Reb Bull "energy drink" is banned in France (and Denmark).

[QUOTE=gbvalor]In addition to the save and interims files, I put some effort doing the files universaly usable. So the itanium intermediate result files can be used by an alpha, a pentimu4 or most of the current popular processors. [/QUOTE]

I can confirm this feature works well. I have swaped between Alpha and Itanium2 systems without any problems.