No, waiting for spring. :)

There's a one in a billion chance that the LL and DC tests will both be wrong saying it's composite when it's really prime so that should help discourage triple checking if we've only tested several million. Thgere we have about 1 in 1000 error rate but if someone thinks up of something it could be even less likely. Also how do you change the speed of Prime95? I don't have to have it idle when I'm not using anything else.

Clowns789

P.S. Where did they get "95" in Prime95?

[quote]Also how do you change the speed of Prime95? I don't have to have it idle when I'm not using anything else.
[/quote]

There's no need to change it coz it won't speedup a test. If you're not using the computer all free cpu cycles will already go to prime95

[quote="clowns789"]P.S. Where did they get "95" in Prime95?[/quote] I believe it is because the program was originally written for Windows 95 several years ago, as opposed to being written for Windows 3.11 or some other such OS.

There's still a link to the Windows 3.1 version (prime.zip) on [url=http://www.mersenne.org/freesoft.htm]http://www.mersenne.org/freesoft.htm[/url]

[quote="http://www.mersenne.org/freesoft.htm"]Windows 3.1: Download prime.zip(202K). This version is not fully automatic, but you can use the PrimeNet server's web-based forms to get exponents and report results. ECM factoring is not supported. Version 18.1, last updated April 2, 1999.[/quote]

[quote="clowns789"]There's a one in a billion chance that the LL and DC tests will both be wrong saying it's composite when it's really prim[/quote]

In fact the historic error rate is a little under 2% which means that less than four exponents in every ten-thousand will return two incorrect results.

The probability that these two results will match is 1 in 2^64, i.e., negligable.

Regards

Daran

[quote="Mr. P-1"][quote="clowns789"]There's a one in a billion chance that the LL and DC tests will both be wrong saying it's composite when it's really prim[/quote]

In fact the historic error rate is a little under 2% which means that less than four exponents in every ten-thousand will return two incorrect results.

The probability that these two results will match is 1 in 2^64, i.e., negligable.

Regards

Daran[/quote]edit: never mind, I re-read your math...

Any word on the status of the new server? Still in the conceptual stage, preliminary work, etc?

Just a quick question for Mr P-1 (or anyone else knowledgeable): Do we know for certain that none of the matching sets of residues are both "common" errors (e.g. residue = 2) caused by hardware errors such as the one which created the M40 false alarm?

Waking up dead thread now... :grin:

[QUOTE=Complex33]Any word on the status of the new server? Still in the conceptual stage, preliminary work, etc?[/QUOTE]

Well I think it's now beyond preliminary work. There's an API, v5.mersenne.org actually points somwhere too!

Something official from Scott would be nice.

Also, I don't know if this board prunes threads, but if so, this thread is fairly important so I ask that one of the mods exclude it.

[QUOTE=PrimeCruncher]Waking up dead thread now... :grin:



Well I think it's now beyond preliminary work. There's an API, v5.mersenne.org actually points somwhere too!

Something official from Scott would be nice.

Also, I don't know if this board prunes threads, but if so, this thread is fairly important so I ask that one of the mods exclude it.[/QUOTE]

This thread has now the sticky attribute. :smile:

Luigi