Doublecheck drive #1: k=3-1001 n=100K-260K

[mdettweiler]

Hi all,

This drive will be run similarly to the first-pass team drives except for a few key differences. First of all, these numbers have already been crunched before, so we will probably find relatively few new primes (because the only primes we will find are ones that were missed the first time around). However, in previous doublechecking efforts for the 300<k<=1001 range, there was a somewhat substantial error rate, so I can practically guarantee that we will find some errors.

Secondly, due to the large number of k's (500 total!), and for the sake of easier checking with first-pass results, we are doing reservations for this drive by k, rather than by n as is normally done in the first-pass drives.

NEW NOTE as of 2010/02/27: We will be posting files with 5 k's in each of them sorted by k-value for the entire range of n=100K-260K. Feel free to reserve as few or as many k's as you wish at a time and pull the applicable portion out of the posted sieved files.

Third, we have to limit this drive exclusively to machines that we are extremely confident in the stability of. If your machine is overclocked, it is required that you run a Prime95/mprime stress test on your machine for at least a few hours to ensure that your machine will produce good results. It is also recommended for non-overclocked machines, though for non-overclocked machines, if you are reasonably confident in your system's stability, and you've run some other form of stability confirmation (such as a GIMPS self-test before doing an exponent, or a few GIMPS, SoB, or PrimeGrid BOINC doublecheck numbers successfully completed without error on your machine's part), you can skip the stress test if you want.

The stringent entry requirements are needed because we don't have any first-pass residuals to compare with (as contrasted with the doublecheck projects mentioned in the previous paragraph), and thus we have no way of checking whether the residuals match--we only have primes found to compare. Thus, we need to have the utmost quality of results for this effort. In future doublechecking efforts, such a requirement will not be needed because we will already have residuals to compare with.

We will have reservations available for the 3-300 and 300-1001 ranges of k separately, for the purpose of variety. The higher k's take slightly longer to test, though there isn't that much difference.

It is not explicitly required that the lowest available k-values be tested due to the varying weights of each k. Although we would ask that the reservations be kept as close to the lowest available k-value as possible.

Below are files available for testing. Feel free to take one k at a time.
This drive is complete--thanks to all who participated! Please see the k<300, n=260K-500K extension where you can find plenty more work of similar nature if you are so inclined.

Reserved and available ranges:

3<=k<300 [COMPLETE]
  k       reserved by    status
----------------------------------
3-13      Mini-Geek      complete
15-19     Flatlander     complete
21-25     kar_bon        complete
27-31     gd_barnes      complete
33        Flatlander     complete
35        Harvey563      complete
37        Flatlander     complete
39-43     gd_barnes      complete
45-49     Harvey563      complete
51-55     gd_barnes      complete
57-61     Harvey563      complete
63-75     gd_barnes      complete
77-81     Harvey563      complete
83-89     gd_barnes      complete
91-111    Harvey563      complete
113       Flatlander     complete
115-159   Harvey563      complete
161-163   grobie         complete
165-177   gd_barnes      complete
179       grobie         complete
181-299   gd_barnes      complete
--------------------------------
300<k<=1001
  k       reserved by    status
----------------------------------
301-307   MyDogBuster    complete
309-311   mdettweiler    complete
313       MooooMoo       complete
315-321   mdettweiler    complete
323-333   Flatlander     complete
335-343   mdettweiler    complete
345-399   gd_barnes      complete
401-405   mdettweiler    complete
407-417   gd_barnes      complete
419-423   mdettweiler    complete
425-499   gd_barnes      complete
501-599   Flatlander     complete
601-699   Sloth          complete
701-799   gd_barnes      complete
801-849   Flatlander     complete
851-1001  Sloth          complete

The following k's <300 have also been double-checked for n=260K-500K. If you would like a sieve file for one or more k's, please email me at max@noprimeleftbehind.net. (Note: doublechecking of k=300-1001 for n>260K will be handled separately, as for those we have first-pass residuals on file from NPLB's testing of that range.)

  k       reserved by     status
----------------------------------
5-15      Harvey563      complete
17        available
19-29     Harvey563      complete
31        gd_barnes      complete
33-299    available

Missing and incorrect primes found:

prime              comments
27*2^282700-1      missing from Rieselprime.org
29*2^347896-1      missing from Rieselprime.org
405*2^104462-1     missing from Primesearch
405*2^109496-1     missing from Primesearch
405*2^193516-1     missing from Primesearch
437*2^138124-1     missing from Rieselprime.org
695*2^103822-1     missing from Rieselprime.org
713*2^111226-1     missing from Rieselprime.org
939*2^149765-1     missing from Rieselprime.org

Please put both primes.log (or whatever you told LLR to use as an output file) and lresults.txt in a zip file at the end of your range and attach the zip file to your post when you report a range as complete. If your results files are too big to attach here, you can email them to me at max@noprimeleftbehind.net, though I'd rather if results were attached here whenever possible.

Let's have fun and plow through those doublechecks!

[TimSorbet]

Taking 9-13.
Edit: If possible, what are the known primes in my range? Oh, and, my CPU isn't overclocked, yes I'm more than reasonably sure my comp is stable, and have run the Prime95 self-test, and have returned a double check.

[TimSorbet]

Taking 3-7 also.

[Flatlander]

Taking 15-19.

btw I've run a torture test for 16 hours without any problems, but I will lower the CPU speed a bit just to be sure.

[mdettweiler]

Originally Posted by Mini-Geek

Taking 9-13.
Edit: If possible, what are the known primes in my range? Oh, and, my CPU isn't overclocked, yes I'm more than reasonably sure my comp is stable, and have run the Prime95 self-test, and have returned a double check.

We haven't checked for known primes yet--I'll let Gary do that.

As for your machine: Okay, sounds good.

[mdettweiler]

Originally Posted by Flatlander

Taking 15-19.

btw I've run a torture test for 16 hours without any problems, but I will lower the CPU speed a bit just to be sure.

If you've run a torture test for 16 hours without a problem, then you shouldn't need to lower the CPU speed, though of course it wouldn't hurt.

[gd_barnes]

Originally Posted by Mini-Geek

Taking 9-13.
Edit: If possible, what are the known primes in my range? Oh, and, my CPU isn't overclocked, yes I'm more than reasonably sure my comp is stable, and have run the Prime95 self-test, and have returned a double check.

Well, guess what I've been doing the last 2-3 days? I've been bringing Karsten's k<300 page up to date and making it non project-specific.

Since Kosmaj is hiding the RPS k<300 page due to the perception of people picking up the testing ranges of k's for their own benefit, something that Karsten, Anon, and I vehemently disagree with, I will make it available for viewing on request in a PM. I still have some links to add but the info. is 100% accurate and up to date as of early this morning.

You'll be able to see the known primes for your k's < 300 in full color. lol


Gary

[gd_barnes]

In suggesting doing this by k, I has assumed that the files would be sorted by k primary and n secondary.

I would now suggest that people wait until they are done with their entire 3k file before posting any primes. We'll then have to sort them by k to make it easy to verify them.

The best alternative is probably for people to attach their primes files here after finishing each file to avoid myriads of primes posts.

Thanks to everyone for hopping in on this. I'll move 3 cores over to it around 4/1 after sieving of k=300-400 for n=600K-1M is complete to P=14G.


Gary

[mdettweiler]

Originally Posted by gd_barnes

In suggesting doing this by k, I has assumed that the files would be sorted by k primary and n secondary.

I would now suggest that people wait until they are done with their entire 3k file before posting any primes. We'll then have to sort them by k to make it easy to verify them.

The best alternative is probably for people to attach their primes files here after finishing each file to avoid myriads of primes posts.

Thanks to everyone for hopping in on this. I'll move 3 cores over to it around 4/1 after sieving of k=300-400 for n=600K-1M is complete to P=14G.


Gary

Yes, I agree--we should post primes files here as well as lresults files. Since there's a limit of one attachment per post, just pop them into a zip file and you'll be all set.

As for sorting by k primary and n secondary, unfortunately srfile can only sort by n and not by k, so thus the files are sorted by n. However, as you were saying, as long as we just post the primes files too at the end of the range, we'll be fine.

[mdettweiler]

I've just added some files for 400<k<1001. Now you all have got a choice of two ranges to reserve from!

BTW: I just ran the mprime torture test on my Core2Duo (the other core's running LLR, so it should be at as full of a load as it's going to get) for an hour, without any errors, so I'd say it's all clear to go for this doublechecking now. I'll hold off on reserving a range, though, because I've got some other reserved work to clean up first.

[gd_barnes]

Originally Posted by Anonymous

Yes, I agree--we should post primes files here as well as lresults files. Since there's a limit of one attachment per post, just pop them into a zip file and you'll be all set.

As for sorting by k primary and n secondary, unfortunately srfile can only sort by n and not by k, so thus the files are sorted by n. However, as you were saying, as long as we just post the primes files too at the end of the range, we'll be fine.

Srfile can write out by k using the -g switch. This is ideal for what we want. You then just do the copy command to copy 3 files at a time into a 3k file for LLRing. There's two extra instances of the header in the file but LLR doesn't care.

Obviously you're sorting by k in some manner. However you got it down to 3k files is the way to get it down to 1k files and then combine them with a copy command into one file.

Excel can also be used for sorting small files like this but would take quite a bit longer with 167 or so total 3k files.

This is just for future reference. No need to mess with it now since people will post files of primes at the end of a range. It's how I divided k=300-1001 for n=50K-100K up into 50k chunks for splitting amongst 7 machines.


Gary