Prime gap records

[mart_r]

The WR gaps so far have all been fully checked, consequently this gap should also be fully checked IMHO.
My pocket calculator says for 36375 tests it would take me about 82 days for a full check on my Win8-PC. Depending on the progress, later I can also allocate some more resources (two more cores on my newest laptop which is a bit faster than the PC). I haven't tried -a1 in OpenPFGW yet.
But for a start it would be good if you could just provide the list of values -3531714 < x < 3654858 for which 29*479909#/19#+x has to be checked (either here or via PM). I would then pick a range, x < -2000000 say, and start testing.

[kruoli]

If open for anyone, I would like to help with testing x > 3e6.

[robert44444uk]

Originally Posted by MJansen

I would be willing to do that, I will run the remaining candidates on 12 cores if you can let me know how far each of the 4 files has progressed. Send me a PM for the details.

Kind regards
Michiel

Michiel, I sent the candidate to you in an email

[MJansen]

Originally Posted by mart_r

The WR gaps so far have all been fully checked, consequently this gap should also be fully checked IMHO.
My pocket calculator says for 36375 tests it would take me about 82 days for a full check on my Win8-PC. Depending on the progress, later I can also allocate some more resources (two more cores on my newest laptop which is a bit faster than the PC). I haven't tried -a1 in OpenPFGW yet.
But for a start it would be good if you could just provide the list of values -3531714 < x < 3654858 for which 29*479909#/19#+x has to be checked (either here or via PM). I would then pick a range, x < -2000000 say, and start testing.

Thanks Mart!
and also thanks in advance to Kruoli!

Regarding logging: using the minus L option: -lyourlogfilenamehere.txt command in the command line, will log all output into a textfile in the directory where the pfgw64 executable file is. I am not sure what -a1 does, I remember Jens asked me to use that in 2013. Logging will do the trick just fine!

Attached you will find the list of candidates after sieving till 10^12. In 2 files, on for the plus side and another for the minus side. It has all candidates in a range 29*Prime(40.000)#/9699690 +/- 50*Prime(40.000). Just delete the extra values outside the range -3531714 < x < 3654858. The first 3 rows look like:
29
9699690
479909

Those are the multiplier (29), the divider (9699690) and the primorial of Prime(40.000) = 479909#
From the 4th line onwards you will find the remaining candidates on the minus or plus side, each side has a separate file.

Note: The 62.5 days was based on around 10 minutes per PRP on my machine. Taking 13 minutes per PRP on yours would indeed take 18 extra days... There should be a smarter way ...

Kind regards
Michiel

[kruoli]

Hm, PFGW 4.0.1 sees my CPU as having only (up to) SSE4.2, while it has AVX512. Is this the wrong version? I took it from SourceForge.

I changed the file header to ABC 29*479909#/9699690+$a because I did not see which format your file is in.

[mart_r]

OK, like I said I'll start testing 29*479909#/19#+x with -3531714 < x <= -2000000.
I won't test those -23995430 <= x < -3531714, that would take way too long

[MJansen]

Originally Posted by kruoli

Hm, PFGW 4.0.1 sees my CPU as having only (up to) SSE4.2, while it has AVX512. Is this the wrong version? I took it from SourceForge.

I changed the file header to ABC 29*479909#/9699690+$a because I did not see which format your file is in.

Hi it is a custom made flat text file, they are a direct output from Jens's Treesieve program. Change them any way you like, ABC seems easy enough.

Thanks again for helping out!

Kind regards
Michiel

[MJansen]

Originally Posted by robert44444uk

Michiel, I sent the candidate to you in an email

Thnx, recieved them in good order and the gapsearch has continued just now, keep you posted!

Kind regards
Michiel

[MJansen]

Originally Posted by mart_r

OK, like I said I'll start testing 29*479909#/19#+x with -3531714 < x <= -2000000.
I won't test those -23995430 <= x < -3531714, that would take way too long

Lol, good thing you did not start the second part ;-)

[SethTro]

Michiel had some excellent suggestions about updating the "primecert" column to more than

"C" - certified
"P" - probabilistic
"?" - unverified

In short they suggest adding, P (proven), D (Double checked), T (Trusted submitter) with upper and lower casing to indicate if it's the endpoints (lowercase) or internals (uppercase) that have been proven / verified / taken on trust.

I've added a new page modern-format to the wiki to formalize this change.

Michiel will propose the new text for P/D/T and everyone else is welcome to suggest other changes.

Hope everyone is well and enjoying the spring; My tulips have just bloomed!

[kruoli]

Originally Posted by kruoli

If open for anyone, I would like to help with testing x > 3e6.

This is finished. See the attached file.