New exponents for TPS

[MooMoo2]

Gary has kindly provided an updated PRPNet server for TPS, which allows us to run newer and faster versions of LLR. Since there'll only be one or two TPS ports, I'd like to determine which candidates we should test. To summarize, I'd like to gauge interest in the following options:

1.) Adding quad-sieved candidates for n=1.7M. Any twins and Sophies found would be record-setting, but non-twins/sophies would not make the top 5000 list.
2.) Adding quad-sieved candidates for n=3.322M. These are a bit over a million decimal digits long, and primes that aren't twin or sophie will make the top 5000 list.
3.) Continuing with twin-sieved candidates for n=480K-500K, k=909K-10M. There's no realistic potential of finding Sophies, and any twins found would currently be the third-largest.

As always, feedback is greatly appreciated. Thanks for your participation!

[rogue]

What does "quad-sieved" mean? Since this is combined with a search for Sophie-Germain primes I assume you are sieving k*2^n+1, k*2^n-1, 2k*2^n+1, and 2k*2^n-1. What sieving program? I'm not suggesting that anything in mtsieve is faster. I'm jus curious as to what sieving program is being used.

[MooMoo2]

Originally Posted by rogue

What does "quad-sieved" mean? Since this is combined with a search for Sophie-Germain primes I assume you are sieving k*2^n+1, k*2^n-1, 2k*2^n+1, and 2k*2^n-1. What sieving program? I'm not suggesting that anything in mtsieve is faster. I'm jus curious as to what sieving program is being used.

"Quad-sieved" means four forms are sieved. I used the "Lucky Minus" option on NewPGen. If k*2^n-1 is prime, k*2^n+1 is checked for twins, and both k*2^(n-1)-1 and k*2^(n+1)-1 are checked for Sophies:

21807225*2^3322-1 is prime! (1008 decimal digits) Time : 25.559 ms.
21807225*2^3322+1 is not prime. Proth RES64: 43E7FB879D8EC522 Time : 14.570 ms.
21807225*2^3321-1 is not prime. LLR Res64: 47D8D22FC652EDF9 Time : 26.763 ms.
21807225*2^3323-1 is not prime. LLR Res64: 7CDE20EA4062A7C9 Time : 39.094 ms.

The first part of the sieve is by far the slowest part. Once you hit p=10T or so, the sieving rate exceeds 100M/sec on one core of a relatively recent PC.

[Bottom Quark]

Originally Posted by rogue

Since this is combined with a search for Sophie-Germain primes I assume you are sieving k*2^n+1, k*2^n-1, 2k*2^n+1, and 2k*2^n-1.

That appears to be a bi-twin search:
https://www.mersenneforum.org/showpo...87&postcount=2

2582509380*2^5090+1 is not prime.  Proth RES64: 1A068571957DB9B8  Time: 245.429 ms.
2582509380*2^5090-1 is prime!  Time : 37.032 ms.
2582509380*2^5091+1 is not prime.  Proth RES64: 0481C51F75EA9C5B  Time: 142.136 ms.
2582509380*2^5091-1 is prime!  Time : 37.098 ms.

[Bottom Quark]

The poll results are attached.

[The Carnivore]

Originally Posted by Bottom Quark

The poll results are attached.

All aboard the meme train!