Distributed Emirp Project

[rsbriggs]

You should at least have a link to emirps on there.

I've kind of dropped out here - got another 5xxxxx block and it was taking about 3 minutes per factor. That's going to tie up a box doing unnecessary work for too long for my taste, so I just shut it down. Be sure to post something when you get some of those optimizations done to eliminate "bad" blocks like that. Unlike some others with big farms, I really don't have very many CPU cycles that I can spare.....

[excaliber]

Quite understandable. In fact, Im thinking about putting the project on hold until I get some much needed improvements done, such as bugfixes and the optimizations.

I think in the long run that will make people happier than using a frustratingly slow client.

3 minutes a number is unacceptable, and needs to be fixed.

[prokaryote]

On the bright side, this is some good info to include in your science fair project (what considerations have to be taken into account in order to successfully implement a DC project)!

prok

[rsbriggs]

I can even justify the long client processing times in the name of showing total processing time versus your single computer time for your overall project.

What I just can't tolerate is watching the program take that long checking the pairs of numbers and their reverse for primeness when the numbers are like the
583064 , 460385 pair. That just grates on my nerves a little too much...

Overall I'd say you have an amazing infrastructure set up, given the resources (just you) allocated to the project. The client even seems to be stable, which even puts you ahead of some of the other "real" DC projects going.

I think you'll need to change the scoring a little to make it more fair - obviously the earlier blocks, like 1-100, shouldn't count for nearly as much as the later blocks.

Team stats are important. If you look at distributed folding, for example, you'll see that the competition between the top couple of teams (FreeDC, ARS) is responsible for a big part of the CPU cycles being dedicated to the project.

People tend to gravitate towards projects they have an interest in, and sometimes the expertise of the "crunchers" can exceed the expertise of the project organizers in some areas. It's good to keep an open dialog going with the crunchers, and you might just learn some things along the way, like AMDs little trick for simultaneously testing multiple small divisors.

And, all of these little discoveries could/should be mentioned as contributing to a successful DC project.....

[cygnussphere]

I don't have the spare boxen to throw on a client like this but I would like to toss in a collective pat on the back and keep up the good work!
We are all now looking forward to see the maturation of your project

[excaliber]

Thanks again to everyone for suggestions, comments and support.

I'll get back to you guys later with a more mature project, with the kinks worked out.

I've also gotten a suggestion from Ars people to consider an encryption project (something simple like 40 bit), which might be considered. Im thinking about it for next year. Who knows, I could even tackle something bigger, like the latest RC5...as cracking one encryption is like another, just longer and more processing (and infrastructure to support it)

Thanks again!

[rsbriggs]

I certainly hope that you aren't going away discouraged here. As far as "maturity" goes, you put together as good or better an infrastructure as I've seen on on many DC projects. Every piece of it was top rate in my book, from the stats, to the forum, to the client.

Throw together some team stats, do some sort of encryption breaking project, then mention that ARS thinks THEY can be the top team, and you'll have more volunteers trying to prove them wrong than you could possibly imagine, and more crunchers than you care to deal with, probably. (We do get cranky and hard to deal with when stats don't work, for example. And there is always someone that thinks they can do it better/faster/more efficiently than you are doing it... )

Seriously - you deserve to be congratulated for what you put together there, it's every bit as good as some projects that have been on-going for years. Don't underestimate yourself - you actually did one heck of a good job with this project.

I'd be more than happy to do some crunching on your re-organized project - keep us informed about what you have going on !!!!

[jonnyw]

well i dont know what's happenend but since the talk a few posts above about dns what-de-ya-ma-callit the site seems to be working for me.

Why it didn't earlier I really don't know ???

Anyway, when the optimisations have been sorted i'll give it a go, but at the moment my complete dc farm consists of 1 pc, which is running distributed folding, so at the moment I have none to spare

[excaliber]

rsbiggs: Not discouraged in the least. I've learned alot of important things that I need to take into consideration.

And, to top it off, the client would be fairly easy to switch to a new project. Its modularized, so I would only have to switch the file manipulation and actual crunch module. In theory

I think i will stick with the emirps a bit longer.

Thanks again.

Jonnyw: Sounds like my 'farm' I have two actually, but the one is so slow it doesnt count

[cygnussphere]

Originally posted by rsbriggs
I certainly hope that you aren't going away discouraged here. As far as "maturity" goes, you put together as good or better an infrastructure as I've seen on on many DC projects. Every piece of it was top rate in my book, from the stats, to the forum, to the client.

Beings as you seem to have taken exception to my choice of words.
Let me clarify.
What I ment by maturation was seeing excaliber's project progress to its end in terms of the high school aspect. How did it work out? Did ya win, Did ya get a good grade kinda stuff....

It was not ment to be critical in any way.

I sincerly hope he didn't take it that way.

[excaliber]

Lol. Nope, didnt take it that way at all.

I figured you meant maturation in the sense of the growth of something. When my projec t matures, it will have grown.

Didnt take it in the least as negative (immature child, etc) :P

[Paratima]

Yeah, excaliber. We're all waiting for your next.

Give Free-DC the first shot at it, OK?

[excaliber]

No problem.

In the idea box are:

n!+/- 1 primeforms
bigger emirps (past compiled prime lists)
crypto

[rsbriggs]

bigger emirps (past compiled prime lists)

You'd have to be certain that your routines are capable of handling multi million digit numbers. The list of known primes goes out quite a long ways, with the largest known now standing at 2 ^ 13466917-1 - which has just over 4 million digits...
Here are the just the first few digits of the number:

924947738006701322247758382547664051925354401079958299021030936080295656580
559610047613121555730584649024542650476541902097669790582767254479290261580
667487156696716687781578148385468287406821390467694278095177459820810948631
467725762962570034225395821145117842917404555309456566569402103759689777345
104671906344606780985345593482044157148165716238608288581716958496154434714
717468371306759548487677488580857871528045254445185145540099077754629508118
775475303098371500921633967355074023207926536138198568216429088393775203400
389763768175752573830192582002581510260270589015776507414369996094315687506
373190139575113739703543485012549547233270163855657410761366571062373138146
723947480164414019101176525165804943863315424962629146699583917451384909987
471792199241089008377215317220839399152562603308218864937648629057952185962
540238050157491687847495951406779716414425617700893144343103374926499454237

[excaliber]

Ahh....

I also got another suggestion. Try to compute Emirps in a different base. Such as base 8 or hex. That way there is something 'useful' going on that cannot be found with a simple pre-compiled prime lookup

For the entire conversation that preceded that, look here (good read too)

http://arstechnica.infopop.net/OpenT...785#8310902785

Once again, thanks for all the help