PDA

View Full Version : A new sieve program? Anyone?



CaptainMooseInc
04-04-2005, 09:03 PM
Hey everyone, you can look over at the Free-DC forum under the Seventeen or Bust (Free-DC) team forum (not this official one) for the whole post. I have a few questions because I'm running an AMD machine and I think sieving is faster for AMDs than p-1ing right? Well, I need some help if I'm going to switch to sieving and need some links to follow so that I can get in the groove. At the moment I'm using the SB.EXE client which everyone tells me is for p-1ing...

I don't care if I find a prime or not, I just wanna get the best use out of my machine. Here is an excerpt from a post of mine:

Which client do I use if I want to sieve? (If it's LLRNET then I'll stick with p-1ing). I wish someone would make a program for sieving that would allow you to reserve a range via the program. The program would download a list of "reserved" and "available" rangers from a host server. If the server has waited too long for a "reserved" set of data to be returned, it would automatically bring it back to the "available" list. Then you could highlight a range (or a few ranges) that you would like to "reserve" and sieve for.

This would make the "babysitting" task less time consuming and you wouldn't have to look through threads to see who's "reserved" what. It would all be server handled and "distributed"...

If someone thinks this is a decent idea and knows how to get into contact with the SoB people easily, then be my guest and pass it along. I really feel too "new" to the project to be throwing ideas into the project managers' faces...

-Jeff

vjs
04-04-2005, 09:46 PM
Thanks for yur interest moose, and yes all of your idea and questions have been discussed before. Sounds like your starting to think about hte project and correctly so if your thinking sieve.

Start by looking at the reservation thread...



To start sieving first visit MikeH's Site then download the cmov client for non-P4's or the sse2 for P4's. Extract the new dat into the same folder as the client. Use Winzip to extract them. tutorial for rieselsieve may help out. For references purposes the old dat, a clear text version, and A very small zip file, indicating the next couple of days of main and DC PRP efforts and the daily shrinkage of the new sob.dat file.
A Excel spreadsheet, to help you calculate, g/day based upon speed, time to complete a range. Also Sobistator, a great addon tool!!!
A complete holes report can be found here


And I'm pretty confident sieve won't go automatic for quite some time it's actually has advantages that way.

I suggest you read through these threads a bit and then ask questions there is really alot to read.

THe best thing to do is reserve a 500G range and start it's all downhill from there.

CaptainMooseInc
04-04-2005, 10:32 PM
See, it's the babysitting that I don't like about sieving.

What would you say the "coverage" of sieving is at the moment? Are there enough/an adequate amount of sievers for those who are PRPing?

If so I'll stay PRPing because it seems like too much work and I don't wanna have to come back and reserve more numbers.

If the speed for AMDs is faster then its benefits aren't good enough to entice me away from PRPing. I tried the QQQsecondpass registry trick and it seemed to move slower than normal PRPing.

Maybe once sieving evolves somewhat, then it will attract more people like me who would love to do it, but the cons outweigh the pros.

-Jeff

hhh
04-05-2005, 12:40 AM
See, it's the babysitting that I don't like about sieving.

Well, to sit a baby is definitely more work than sieving, though.:)

Once you got used to the process, I think, you can choose ranges so that you have to take care less then once a week.
(I am factoring, which is said to be more work, and it is possible).
But yes you are right, it's not fully automated.
As for the need for sievers, you are welcome, espescially with the new sob.dat which sieves up to 50M, there is enough work for everybody. You find information in the other threads.

Espescially in the beginning, it might be more work, and if you would like to spare your time, but help the project a little bit different then most people, Doublecheck might be your choice. It seems to be slower, but this is a problem of statistics (ECM/s are a bad measure for the work done, small n give you less credit), and your computer is working at the same speed.
As DC is way behind the main effort, a little help is welcome, too.

See you, somewhere... H.

vjs
04-05-2005, 09:03 AM
IMHO,

Both sieving and factoring are way underpowered... we should be at a sieve point where it take the same amount of time to sieve for the factor found as it does to test those numbers. If not 1.5X the time it take to test those numbers. This would probably mean we should be sieving somewhere areound 2000T or two to three times as deep at least. (And yes we should be using 991<n<50M since it's near the optimal dat size.)

As for babysitting??? Well you don't have to babysit at all. You could reserve 2 months worth of sieving and submit factors every two months, in less time than it took you to post...

Also, hhh, is correct about second pass. Regardless do as you wish, how fast is you machine or machines?

CaptainMooseInc
04-05-2005, 01:26 PM
Running an AMD Sempron 2800+ (2.0GHz) w/ 512 MB PC2700DDR, WinXP Pro.

-Jeff

CaptainMooseInc
04-05-2005, 01:27 PM
I do have a 2.7GHz Celeron that is crunching too, I just noticed that my AMD Sempron is PRPing faster than the Celeron. :)

-Jeff