Discussion about CPU time needed and k's remaining

[KEP]

@Gary:

You're all forgiven... I see now, that I should also start getting a bit better at explaining things, but I tends to wanna tell more than I can reasonably explain such as everyone understands... If you would like the sieved file, I wouldn't mind sending you the srsieve.out file this friday, all I need is to know if you're interested in the sieve file (which I understood you were) or not... Also the Prime Files for Riesel Base 3 is approximately uncompressed 4 GB, so how would you like me to send this data to you?

Regards

KEP

[masser]

Originally Posted by gd_barnes

I had no clue that you had all of these different ranges that could be tested. I thought you just had the two servers, each currently testing somewhere between n=200K and 300K, one of which I worked on for ~7-10 days. Where are these very small and very large n-ranges that can be tested? But I am confused. If there are small n-ranges to be tested, why is that? Wouldn't it better to knock those out and remove huge #'s of candidates from the huge dat file (if a prime is found) if your goal is to ultimately prove the conjectures? If the goal is to prove the conjectures, I think that small n-ranges should all be tested. But perhaps you've determined that the fun that it makes for everyone is more valuable than the proof. I agree there that it's not always easy to strike a balance between the two.

Gary

Those ranges I described are not organized searches, just ranges available in the dat file if someone was interested in testing them. For instance, our double-checking efforts have been totally voluntary. Occasionally someone wants to contribute a slow machine to our efforts, so they test a range from our dat file that has already been tested and get credit on our stats page for double-checking. All of the small ranges (n<260000) have been tested at least once.

[gd_barnes]

Originally Posted by KEP

@Gary:

You're all forgiven... I see now, that I should also start getting a bit better at explaining things, but I tends to wanna tell more than I can reasonably explain such as everyone understands... If you would like the sieved file, I wouldn't mind sending you the srsieve.out file this friday, all I need is to know if you're interested in the sieve file (which I understood you were) or not... Also the Prime Files for Riesel Base 3 is approximately uncompressed 4 GB, so how would you like me to send this data to you?

Regards

KEP

OK, quick response so no miscommunication: Only 1-2 word answers needed:

1. What k-ranges and n-ranges have you searched Riesel base 3?

2. Is the sieve file you are referring to Sierp base 19 or Riesel base 3?

3. Are you able and OK with sending us a sieved file on Sierp base 19?


If you can answer me those 3 questions, I can give specific answers.


Thanks,
Gary

[gd_barnes]

Originally Posted by masser

Those ranges I described are not organized searches, just ranges available in the dat file if someone was interested in testing them. For instance, our double-checking efforts have been totally voluntary. Occasionally someone wants to contribute a slow machine to our efforts, so they test a range from our dat file that has already been tested and get credit on our stats page for double-checking. All of the small ranges (n<260000) have been tested at least once.

Ah, very good idea!

On the higher n-ranges that you were talking about, how does that work? Do you allow people to just pick out any high range (i.e. much > then current testing limits) out of the .dat file that they want to test just for grins? That would be a novel idea. It certainly wouldn't be mathematically efficient but from a project interest perspective for people of all kinds of prime-searching tastes, very interesting! I would not have considered such an idea but may consider it here if you are doing it and you are getting some response to it.

Thanks for providing the particulars about what has worked for your project.


Gary

[KEP]

Originally Posted by gd_barnes

OK, quick response so no miscommunication: Only 1-2 word answers needed:

1. What k-ranges and n-ranges have you searched Riesel base 3?

2. Is the sieve file you are referring to Sierp base 19 or Riesel base 3?

3. Are you able and OK with sending us a sieved file on Sierp base 19?


If you can answer me those 3 questions, I can give specific answers.


Thanks,
Gary

Great Idea with short replys in order to keep communication straight, here is the proper answers:

ad 1. k<=500,000,000 has been searched to n<=25,000 when I report it (in 1 to 2 weeks)

ad 2. The sieve file is for Sierp base 19

ad 3. This friday I'm able to send you the sieved file for Sierp base 19 and I'm totally OK with it.

Regarding ad 1. the k-range and its primefiles is ~4,000,000,000 bytes uncompressed.

Looking forward to your answers. Do you want me to remove the k's already primed for Sierp base 19, before sending you the sieve file?

Kenneth!

[gd_barnes]

Originally Posted by KEP

Great Idea with short replys in order to keep communication straight, here is the proper answers:

ad 1. k<=500,000,000 has been searched to n<=25,000 when I report it (in 1 to 2 weeks)

ad 2. The sieve file is for Sierp base 19

ad 3. This friday I'm able to send you the sieved file for Sierp base 19 and I'm totally OK with it.

Regarding ad 1. the k-range and its primefiles is ~4,000,000,000 bytes uncompressed.

Looking forward to your answers. Do you want me to remove the k's already primed for Sierp base 19, before sending you the sieve file?

Kenneth!

Quick question again:

1. Do you still plan to take Riesel base 3 to k=500M and n=25K? I wasn't clear after reading your Email.

Quick answers:

1. On Riesel base 3, if possible, please sort the file by n-value and send me only the primes for n>1000 or n>2500.

2. and 3. Please send me the Sierp base 19 file whenever you are ready. Please also include the sieve limit or checkpoint file. Thanks.

4. Yes, please remove the k's with primes from your Sierp base 19 sieve file.


On #1 answer, I can easily recreate the primes file up to n=1000 or n=2500. There is no reason to send mass amounts of small primes around.


Gary

[KEP]

Originally Posted by gd_barnes

Quick question again:

1. Do you still plan to take Riesel base 3 to k=500M and n=25K? I wasn't clear after reading your Email.

Quick answers:

1. On Riesel base 3, if possible, please sort the file by n-value and send me only the primes for n>1000 or n>2500.

2. and 3. Please send me the Sierp base 19 file whenever you are ready. Please also include the sieve limit or checkpoint file. Thanks.

4. Yes, please remove the k's with primes from your Sierp base 19 sieve file.


On #1 answer, I can easily recreate the primes file up to n=1000 or n=2500. There is no reason to send mass amounts of small primes around.


Gary

1. I'll send you the primes with n>500 (is this OK, since it will be at most 300000) They will be sorted by growing n.

2. I'll send you both the checkpoint file, if any (since I used srsieve to sieve the k's)

4. I'll remove the k's before sending you the sieve file

I'm at the moment taking the k<=500M range for Riesel base 3 to n<=25,000. However due to the manual intervention for the PRP being composites for n<=500 it is going to take some weeks before we can say for certain which k's is remaining.

Regards

KEP

[gd_barnes]

Originally Posted by KEP

1. I'll send you the primes with n>500 (is this OK, since it will be at most 300000) They will be sorted by growing n.

2. I'll send you both the checkpoint file, if any (since I used srsieve to sieve the k's)

4. I'll remove the k's before sending you the sieve file

I'm at the moment taking the k<=500M range for Riesel base 3 to n<=25,000. However due to the manual intervention for the PRP being composites for n<=500 it is going to take some weeks before we can say for certain which k's is remaining.

Regards

KEP

1. Great! Getting all the primes for n>500 works well and the file is not too big.

That also sounds good on #2 and #4.


Question:

Can you post here 3 or more examples where a PRP is composite? It sounds like you have a lot of them but they are supposed to be extremely rare.


Thanks,
Gary

[KEP]

Originally Posted by gd_barnes

1. Great! Getting all the primes for n>500 works well and the file is not too big.

That also sounds good on #2 and #4.


Question:

Can you post here 3 or more examples where a PRP is composite? It sounds like you have a lot of them but they are supposed to be extremely rare.


Thanks,
Gary

Well there isn't many, but it means that I have to search through 2 log files of a total of 50 GB... my personal guess is that there is between 1500-2500 out of a range of 498 million k's. But of course as soon as I find the first k/n PRP pair being a composite, I can post it to you, or if you like any more I can post more as well

@Michaf: (or others who can code some scripts)... would it be possible for you to make a perl script which can do the following:

1. Use the primes.out as input file (for verification process only)
2. Write the composites to a noprimes_PRP.out file (or something similar)

Regarding number 2, if the script can cature those PRP turning out to be composites during verification, it would save one the pressure of having to search through 100 GB log files per 1 G k's. Also the speed that PFGW can run with will be higher in the end, since it takes longer and longer to write to these huge log-files So any chance someone can cook up something that can work for both #1 and #2... if someone can, I really feel like taking the Riesel Base 3 beyond the 500M k's, since then almost no manual intevention is needed

Regards

KEP

Ps. To clarify, I'm not reserving new work, only asking if such programming is possible