Should I be sieving vs prping

[tqft]

OK - my PII 400 MHz, is currently crunching away at about 47.5KcEM/S and slowing.

Have lots of memory and disk space.

Machine basically only gets used maybe 3 times during work week for 1 to 2 hours at night, plus some more on weekends. Is only connected to the 'net when used.

Would it be more useful to be sieving?

I have followed the sieving thread but while all the ideas and knowledge is there - a summary would be very useful. I have the general idea and I think it would be better if my machine was sieving, but there are no easy to read instructions I can find.

Please correct below where I am wrong:
Procedure seems to be - download the sieve client (which client - I think there at least 2 attached in the sieving thread) at http://www.free-dc.org/forum/showthread.php?s=????????,
reserve range of numbers by adding message in thread http://www.free-dc.org/forum/showthr...&threadid=2406
let sieve client loose

Does the client automatically report or have instructions on how to report the results of sieving?

I think correcting the above procedure and then publishing in a standalone thread or on a prominet SoB page would help.

[Joe O]

If you can PRP I would continue to do so as long as your patience holds out. The sieving effort is important, but the primary goal of this project is to find 12 more primes! The sieving effort helps by eliminating candidates from the PRP process. As long as it eliminates them more efficiently than PRPing, or at least as efficiently, it will and should be pursued.

Having said that, if you are still interested in sieving, I will try to gather the information here for you. Check back to this post, as I will edit it instead of posting again when I get more info gathered.

There are currently two sieving efforts. The first is 3M < N < 20M, and the second is 101 < N < 3M. They use the same programs but different SOB.DAT files.

The programs and the 3M < N < 20M SOB.DAT can be found on this page, with an updated program in this thread. The coordination thread for this primary sieving effort can be found here, in this thread. This is where most of the sieving effort is concentrated.

In the unlikely event that we will have to go back and double check our efforts, there is a second sieving effort preparing the way. The SOB.DAT file for this 101 < N < 3M double checking effort can be found in this coordination thread, with a discussion in this thread.

[Nuri]

Although Joe has done a good job of explaining the procedures, there are a few lines I'd like to add that I think might help you tqft.

Considering the configuration, I agree with you in that your machine would be more useful to the project when used in sieving instead of prping. Still, this does not mean that you should sieve.

I believe it depends on your personal preferences. Just try the sieve client with a small range first (I don't know, but I guess a range of 10G would give you the flavour). If you like it, you can reserve another range and continue. If you don't like it, you can simply go back to prp testing easily.

In case you have a problem with Joe's link, here is a direct link to the download file. It's the same program, the difference is this is a direct link to the download.

When you download the zip file and extract it, you will see two exe files. Sobsieveconsole.exe is a few percent faster, but sobsieve.exe is easier to use.

Then, you should decide which sieve to participate. As Joe did, I also recommend you the main sieve. If you decide to do so, download the sieve122.zip file from the sieve page Joe mentioned, and extract the SoB.dat file only. The Sobsieve.exe under this zip file is version 1.22, and is much slower than the latest version (v1.28).

Put the client and sob.dat under the same directory, reserve a range from the coordination thread, and start the client. It doesn't matter which client you choose, they both will ask for the range you want to sieve at the very beginning. Write down the lower and upper bounds of the range you reserved (in terms of billions, as explained in the coordination thread), and start the client.

This will automatically create a sobstatus.dat file in the same folder with your exe and sob.dat files. You will be able to follow the progress very easily in both clients. When your range is finished (or more frequently if you like), copy the factors you found, and submit them through the sieve submission page.

The factors are stored under the sobstatus.dat file. If you are using sobsieve.exe, you can simply copy the factors from the window of the client, and paste at the submission window.

And a last note. You can easily submit factors without logging in to the seventeenorbust site, but then they will be submitted by an anonymous user . If you want the server to know that it's you who submitted those factors, you should log in to the site first.

[Mystwalker]

I don't know, but I guess a range of 10G would give you the flavour

That PC needs approx. 4 hours/G. So it should take ~5 weeks to sieve a 10G range, according to your using profile.
Maybe a 5G range would be better. So you can faster return to prp testing in case you don't like sieving. Hey, most of us started with 5G ranges.

Sobsieveconsole.exe is a few percent faster, but sobsieve.exe is easier to use.

In recent versions, there is not much of a speed difference left - when you fine tune the alpha value of the GUI version, it's sometimes even a tad faster.

[MikeH]

That PC needs approx. 4 hours/G. So it should take ~5 weeks to sieve a 10G range, according to your using profile.

4 * 10 = 40 hours, or just under 2 days!

For that profile of PC, I'd say it's more like 6 hours/G, but still some where between 2 and 3 days, not ~5 weeks.

[Mystwalker]

Machine basically only gets used maybe 3 times during work week for 1 to 2 hours at night, plus some more on weekends. Is only connected to the 'net when used.

This is where I approximated the using (not crunching) profile from:
In average 4.5 hours within the week, I arbitrarily chose 3.5 hours at the weekend (to get a clean result), which would make 8 hours/week usage...


edit:
Wait a moment... Maybe I got things wrong here. The question is: Is that PC turned on when it is not used? If yes, you're of course completely right, MikeH.

[Troodon]

Originally posted by Mystwalker
In recent versions, there is not much of a speed difference left - when you fine tune the alpha value of the GUI version, it's sometimes even a tad faster.

On my machine, the GUI version is faster in Windows 98 SE, while the console version is faster in Windows 2000. Anyway, test both and use the one is faster on your computer or the one you like more .

[tqft]

machine is always on - one of the 3 reasons (not the most important though) I bought it was to run some type of DC project just for fun. Was seti@home but got irritated.

320Mb of RAM (after taking out the mem used by the onboard sound and video - currently 183Mb free physical mem - sob & phoenix brower running + Win98SE)

avergae from stats page @sob is 45kcEM/s

about 5 days to go until current test
k= 28433, n=3499945 is finished (is at 53.5% finished - they are taking about 10 days each)

Have lots of patience to run prp tests and is fun to my slightly warped mind. It's the theory that interests me and seeing it in action is mildly fascinating. Which url's have a description of the theory of the prp test, I have seen mention of the test client using FFT's? Have dug some but hit roadblocks (eg expensive signup to professional journals).

At this stage I may do at least one more test so I can find time to take the procedures below, test them and make sure everything works as advertised. Not nasty criticism, but I have written user & references guides before and I know how easily users can misinterpret anything.

[Nuri]

Originally posted by tqft
Not nasty criticism, but I have written user & references guides before and I know how easily users can misinterpret anything.

I know. Today, for example (and I mean it, it really happened "today"), I noticed that there were two different message boards in on of the forums I've been a regular participant for more than a year.

Anyway, such things do happen. That's why I wrote an alternative post. If you misinterpret Joe's post, refer to mine. If you misinterpret my post, refer to his. If you misinterpret both, please post again, so that we can figure out what has gone wrong.

[Joe O]

Originally posted by tqft
At this stage I may do at least one more test so I can find time to take the procedures below, test them and make sure everything works as advertised. Not nasty criticism, but I have written user & references guides before and I know how easily users can misinterpret anything.

We have an unique opportunity here. TQFT why don't you write an introduction to sieving based on your experiences?

[tqft]

>TQFT why don't you write an introduction >to sieving based on your experiences?

That was roughly the plan