mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Software (https://www.mersenneforum.org/forumdisplay.php?f=10)
-   -   LLRP4 Version 3.3 now available ! (https://www.mersenneforum.org/showthread.php?t=3356)

Jean Penné 2004-11-30 17:02

LLRP4 Version 3.3 now available !
 
Hi All,

The Version 3.3 of the LLRP4 (SSE2 only) program is now available on
the Gimps directory ([url]http://www.mersenne.org/gimps/[/url]), and replaces the
previous one.

The two new features included in this version are :

- Primality proving of k*2^n+1 numbers.
- Primality proving of Carol and Kynea numbers.

Also, some secondary bugs have been fixed...

Thanks to the new George Woltman's code, and to "gwstartnextfft" using,
I observed a speed increase of until 20% for fft lengthes greater than
64K...

More details can be found in the Readme.text included in the compressed
files.

Good primes hunting, and Best Regards,

Jean

em99010pepe 2004-11-30 18:09

Hi Jean,

What's the difference between LLRP4 and PRP3? I tested both and got the same speed. (Newbie question)

Thanks in advance,

Carlos

Pconfig 2004-11-30 18:44

A more noob question. What is LLRP for? :D

Jean Penné 2004-11-30 19:43

LLR, LLRP4 Programs
 
PRP, PRP3 are programs testing if a number is probably prime, by using the
little Fermat theorem (and, indeed, they can assert that a number is composite, if the test is negative).

LLR, and LLRP4 are primality proving programs, that is to say, deterministic.
They are as fast as PRP, PRP3 thanks to using the same George Woltman's code to do fast multiplies and/or squaring of large integers.

Actually, LLR (Version 2.2) can only test k*2^n-1 numbers.
LLRP4, Version 3, can now test k*2^n-1 and k*2^n+1 numbers, but works
only on P4 SSE2 machines (this is the meaning of the "P4").

The next version of LLR (2.3) will also test k*2^n+1 numbers, and will be
released soon, when tested enough.

Regards,
Jean

TTn 2004-12-01 01:27

Bahhh humbug 3.3
 
If the "VerifyUpToLine=x" is still in the ini file, it skips new numbers.
I consider this a bug, since it should still perform a test on the new number, rather than skip it.

Please let me know if this will be changed, so I can adjust RMA.
Since this bug TOTALLY disables the current version.

These curious obsure bugs seem to continue to affect RMA... hmmm.

Jean Penné 2004-12-01 07:11

Nobody is forced to use "VerifyUptoLine" and its default value is 0 (so, not active).
I included this feature to allow persons who were affected by the LLRP41/42 bug to
save time, instead of having to begin all their tests again...

I wish to suppress this option when it will not be used more, but how can I know that ?

Regards,
Jean

BotXXX 2004-12-01 08:59

[QUOTE=Jean Penné]The two new features included in this version are :

- Primality proving of k*2^n+1 numbers.
- Primality proving of Carol and Kynea numbers.
[/QUOTE]

Jean,
This is good news, i am curious how fast the primality proving of Carol and Kynea numbers will be. Currently using OpenPFGW to test these out and now a test run on a little spare machine to try the new LLR out.

TTn 2004-12-01 09:22

I must not be understanding fully.
The behaviour has cleary changed since the first Verify.
As I said before this, I didn't notice any difference with verify on/off.

[QUOTE] I wish to suppress this option when it will not be used more, but how can I know that ?[/QUOTE]
When on, the lresults file could be checked for a matching "new string", if not then do test.

This is the one option(since it didnt work anyway before) that I included by default as checked.
So anyone who downloaded RMA 1.7 beta, please download the newest version that simply allows the option now. No big deal.

Jean Penné 2004-12-29 21:25

LLR Version 3.5 Beta now available !
 
Hi All,

LLR Version 3.5 Beta is now available on the Gimps directory.

This version uses the new Gwnums 24.6 library, an can work on all PC's.

It would obsolete all the previous versions of the LLR program :

- It can prove the primality of k*2^n-1 numbers (with k < 2^n) by using
the Lucas-Lehmer-Riesel algorithm.

- It can prove the primality of k*2^n+1 numbers (with k < 2^n) by using
the Proth theorem algorithm (so it is normal that composite such numbers
yield different residues than those obtained with PRP3 program, which uses
the little Fermat theorem algorithm).

More details can be found in the attached Readme file.

I whish you an happy new year with many new prime discoveries !

Jean

Citrix 2005-01-11 18:46

Jean,
Where is the code for LLR3.5 be included in the LLRnet package. so that more people can just use LLRnet instead of going through manual file reservation etc for various projects etc.

Thanks,
Citrix
:cool: :cool: :cool:

Jean Penné 2005-01-14 07:23

Open Beta Release LLRNet 3.5
 
[QUOTE=Citrix]Jean,
Where is the code for LLR3.5 be included in the LLRnet package. so that more people can just use LLRnet instead of going through manual file reservation etc for various projects etc.

Thanks,
Citrix
:cool: :cool: :cool:[/QUOTE]

Vincent (my son) hast just done this work !
This new LLRNet package can be dowloaded from RieselSieve project Forum,

URL : [url]http://rieselsieve.com[/url]

thread : Forum Index -> Announcement

topic : Open Beta Release LLRNet 3.5

Regards,

Jean


All times are UTC. The time now is 01:24.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.