Prime95 version 25.6

[Prime95]

To al LL testers. I've changed the way LL tests are allocated. The old code would assign exponents that needed no P-1 or TF regardless of the size of the exponent. Now if a much smaller exponent needing P-1 or TF is found then the smaller exponent is assigned.

For those that got exponents in the 50, 60, or even 70 millions, feel free to unreserve them if you do not want such a lengthy assignment.

[philmoore]

Hi George,

Is this feature documented anywhere?

You could try version 25.6 of prime95/mprime. It can PRP numbers of the form (k*2^n+c)/f. Where k is up to about 50 bits, c is positive or negative and less than 20 bits, f is the known factors.

I assume I use 25.6 unconnected to the server and insert something in the worktodo file, but just exactly what?

[Prime95]

Originally Posted by philmoore

Is this feature documented anywhere?

I assume I use 25.6 unconnected to the server and insert something in the worktodo file, but just exactly what?

PRP=k,b,n,c[,how_far_factored,tests_saved][,quoted_comma_separated_list_of_known_factors]

The how_far_factored and tests_saved is used for P-1 factoring.

This feature is not well tested.

[mdettweiler]

Originally Posted by Prime95

PRP=k,b,n,c[,how_far_factored,tests_saved][,quoted_comma_separated_list_of_known_factors]

The how_far_factored and tests_saved is used for P-1 factoring.

This feature is not well tested.

I presume that the basic PRP testing code is the same as is used in PRP v3.0 (and LLR, which uses PRP's code for non-k*2^n+-1 numbers)--i.e. no difference in speed compared to PRP? An informal test I ran a while back showed comparable results.

[nibble4bits]

Is there a plan for a single-core Athlon 64 version for Linuxes?
I have a CPU just waiting to get a client that works on Ubuntu Hardy Heron. Good luck on next prime, Joe

[Prime95]

Originally Posted by nibble4bits

Is there a plan for a single-core Athlon 64 version for Linuxes?
I have a CPU just waiting to get a client that works on Ubuntu Hardy Heron.

Anything wrong with the existing Linux 64-bit port of version 25.6?

[nibble4bits]

Versions listed at mersenne.org:
Windows Vista
Windows 95/98/Me/NT/2000/XP
Windows 64-bit
Linux
FreeBSD
Windows NT/2000/XP Service
OS/2
Windows 3.1

There's no mention of the 64-bit version of Linux. I must have been tired because I can't remember where I saw a 64-bit version and it complained because I have only one core. :/ Maybe the other programmer's version: Mlucas_AMD64_11.03.04.tgz?

I'm looking into my browser history to figure out where the heck it was. LOL Sorry if I mistook yours for another version.

[Uncwilly]

Originally Posted by nibble4bits

Versions listed at mersenne.org:
Windows Vista
Windows 95/98/Me/NT/2000/XP
Windows 64-bit
Linux
FreeBSD
Windows NT/2000/XP Service
OS/2
Windows 3.1

There's no mention of the 64-bit version of Linux. I must have been tired because I can't remember where I saw a 64-bit version and it complained because I have only one core. :/ Maybe the other programmer's version: Mlucas_AMD64_11.03.04.tgz?

Originally Posted by Prime95

Links to 25.6 versions
---------------------
Windows 32-bit - ftp://mersenne.org/gimps/p95v256.zip
Linux - ftp://mersenne/gimps/mprime256.tar.gz
Linux 64-bit - ftp://mersenne.org/gimps/mprime256-linux64.tar.gz
Windows 64-bit - ftp://mersenne.org/gimps/p64v256.zip
Mac OS X - ftp://mersenne/gimps/mprime256-MacOSX.tar.gz

The links to v25.x are not on the website, as they are not yet official releases yet. The links to v25.6 in the first post of this thread.

[nibble4bits]

Oh wow, I meant this:

http://www.stanford.edu/group/pandeg...eta1-Linux.tgz

I guess Standford's FAH's more than a slight leap from Prime95. Oops.

Thanks for the help, Joe

[philmoore]

I am using version 25.6 unconnected to the server, so this may be slightly off-topic to other users, but I have noted that when I edit the worktodo file, I can apparently add lines successfully, but then when the program finishes its current assignment, worktodo reverts to its previous version.

[Prime95]

Originally Posted by philmoore

I am using version 25.6 unconnected to the server, so this may be slightly off-topic to other users, but I have noted that when I edit the worktodo file, I can apparently add lines successfully, but then when the program finishes its current assignment, worktodo reverts to its previous version.

Here's a new undocumented feature. Create a file called worktodo.add to put enter your new work. This should get sucked into your worktodo.txt file at some point. The fact that editing worktodo.txt/worktodo.ini has worked in the past has been a bit of a fluke.