source code?

[shifted]

Is there any chance that a fellow might be able to get the source code (without extensive x86 optimisations)?

[MAD-ness]

shifted: they keep the source code private for security reasons, though the web page states, I think, that if someone has a valid reason for wanting to look at it, they might share it under NDA.

If you are talking about getting the code for a potential port, that is a different kind of thing than just wanting to look at it out of curiosity.

Based on the other thread you started, I assume you want it for a possible port to non-x86 platforms?

[shifted]

Yep!

I have access to several sun machines, an irix machine, and a mac os x machine... However, i'd only be compiling source... i don't know the internals of sparc/mips/powerpc :/

[nuutti]

I think that it is quite big project to build seventeen or bust client to sun because glucas does Lucas-Lehmer test and we are doing
Euler’s probable primality test (I guess). so part of the mathematical code is different in these two projects. I think that mathematical libraries used in sb client are written in assembler so they are not portable at all.

Yours,

Nuutti

[shifted]

Originally posted by nuutti
I think that it is quite big project to build seventeen or bust client to sun because glucas does Lucas-Lehmer test and we are doing
Euler’s probable primality test (I guess). so part of the mathematical code is different in these two projects. I think that mathematical libraries used in sb client are written in assembler so they are not portable at all.

Yours,

Nuutti

Hmm... so there are actually two algorithms involved in the sbclient then?

Could jjjjL elaborate? I am interested in writing a port, if it wouldn't take too much time. I'd rather have sob running than distributedfolding.com's client (better stats! haha)

[MAD-ness]

BTW, shifted, have you looked at Dyyryath's stats for Distributedfolding?

http://stats.zerothelement.com/cgi-bin/distributed-folding/render-teams.pl

That only covers teams, but they have a ton of features (and more being added all the time).

[shifted]

Originally posted by MAD-ness
BTW, shifted, have you looked at Dyyryath's stats for Distributedfolding?

http://stats.zerothelement.com/cgi-bin/distributed-folding/render-teams.pl

That only covers teams, but they have a ton of features (and more being added all the time).

Yikes! That is nice... i wonder how he is getting the data... does he run the distributedfolding site? I'll have to grep around for some answers...

[MAD-ness]

He is just a stats guru.

There are others who run customized stats for Distributed Computing projects.

So far I know of none for Seventeen or Bust, but the stats are pretty darn good that they provide by default anyways.

In the case of DF, when the project stats server updates the stats page, it creates a text file with all of the stats info and it is posted to the page automatically. The stats page makers have scripts to download the file and then update the stats that they display.

www.statsman.org is a place for great stats, for many projects.

http://team.macnn.com/dfold/ you can find links to thier other stats pages there as well.

[shifted]

Sweet :)

[jjjjL]

porting the assembly routines of SB is no easy task... if it were, i would have done it already ;).

i estimate that it would take an expert at least a few months to do it correctly.

if you are interested in trying, grab the code from www.mersenne.org/gimps/source22.zip and port the asm files to whatever platform you can.

if you port those routines (correctly), i'll release a non-x86 client.

good luck.


-Louie

[shifted]

Originally posted by jjjjL
porting the assembly routines of SB is no easy task... if it were, i would have done it already ;).

i estimate that it would take an expert at least a few months to do it correctly.

if you are interested in trying, grab the code from www.mersenne.org/gimps/source22.zip and port the asm files to whatever platform you can.

if you port those routines (correctly), i'll release a non-x86 client.


Months? Yikes. I'll see. No guarantees!

[shifted]

Okay, 500 KB of asm is far too much for me to handle... It would take me weeks to decode it back into a higher level, then weeks to code it back into asm on another machine... hmm...

I guess it would be faster to reimplement the necessary algorithms in C.

I think this will wait until i do my master's, unfortunately :(

[rogue]

Instead of writing your own FFT or reverse engineering GW's x86 source, you could look at YEAFFT (the FFT behind Glucas, the non-x86 GIMPS client) or a few other FFTs freely available (under the GPL). YEAFFT is written mainly in C, so it is very portable.