transitioning to faster LLR

[wfgarnett3]

Hi all,

I just contacted all manual and automated users. Everyone will be transitioning to the new fast LLR!! This will give our project a HUGE speedup. I can't wait :)

P.S.--Thank you George and Jean for incorporated the IBDWT for Proth numbers. Especially George, since he stopped making new Prime95 changes to help all of us (i.e. Sob, Rieselsieve), because now with everyone using IBDWT, any enhancements Prime95 makes we can get too :) :)

Regards,
William

[Jean Penné]

Hi All,

The Version 3.6 of LLR is now available on the Gimps directory!

I tested it on all the verified Proth and Riesel primes from the Chris Caldwell's
database, and got neither false negative nor error message, so it is no more a
Beta version.

It uses the version 24.11 of George Woltman's Gwnums library, which so
demonstrates to be very reliable, many thanks to George!

There are also some new features in this version :
-The k*b^n±1 numbers, were the base b is a power of two, are now
converted into base two numbers before beeing processed, instead of
doing a PRP test on them.

-An iteration in the "Computing U0" loop is two times more time-consuming
than a computing power one, so it is better to make a previous PRP test
before a LLR one, as soon as the k multiplier's bit length reaches 10% of
the bit length of the number to test. That is that is done in this version.
On the other hand, the Proth test is slightly faster than the PRP one, so,
it is always done directly.

-To avoid any confusion, the output line for composite numbers now shows
of which algorithm the residue is the result :

RES64: "xxxxxxxxxxxxxxxx" (PRP test)
LLR RES64: "xxxxxxxxxxxxxxxx" (LLR test)
Proth RES64: "xxxxxxxxxxxxxxxx" (Proth test)

More infos are in the Readme.txt file.

Have fun with this program, and Best Regards,
Jean