I have just submitted to Chris's database a pair of primes that, at 25055 digits, will come in at rank 16 on his Top-20 Twins page
I thought it would be instructive to record the processing steps needed for this quite unremarkable achievement, and thereby maybe encourage others to more ambitious goals in this direction.
The twins are these two:-
Primality testing 1035928263*2^83200-1 [N-1/N+1, Brillhart-Lehmer-Selfridge]
Calling N+1 BLS with factored part 100.00% and helper 0.01% (300.01% proof)
1035928263*2^83200-1 is prime! (124.2008s+0.0011s)
Primality testing 1035928263*2^83200+1 [N-1/N+1, Brillhart-Lehmer-Selfridge]
Calling N-1 BLS with factored part 100.00% and helper 0.01% (300.02% proof)
1035928263*2^83200+1 is prime! (79.9587s+0.0016s)
The first step was to ask NewPGen to sieve a block of numbers of the form k*2^n+/-1, with n=83200 and k=1..4*10^9, to a depth of 179 trillion.
This took 28.4 GHz-days, and left a mere 1,1551,319 candidates.
The next step was to feed the output file to PFGW, which is aware of the NewPGen output format for Twins, and tests the lower one of each pair, and only if that is a PRP goes on to test the higher one.
PFGW found the twin after 120 GHz-days, when it was about 1/4 of the way through the complete NewPGen output file.
Finally, one just needs to give PFGW the 2 PRPs with the -tc processing option, to confirm that they are indeed prime (giving the output quoted above).