switched to AP27 after tdp ended and found 2 AP20s in the last 4 hours. (These are the first APs I've found. Thanks to PrimeGrid Admins or nVidia developers, whoever fixed it so the GPU actually works without returning 100 errors.)
____________

switched to AP27 after tdp ended and found 2 AP20s in the last 4 hours. (These are the first APs I've found. Thanks to PrimeGrid Admins or nVidia developers, whoever fixed it so the GPU actually works without returning 100 errors.)

switched to AP27 after tdp ended and found 2 AP20s in the last 4 hours. (These are the first APs I've found. Thanks to PrimeGrid Admins or nVidia developers, whoever fixed it so the GPU actually works without returning 100 errors.)

Yup there seems to be ALOT of AP20's left to find, I was crunching those for a bit and was finding several almost every day! Every once in awhile I'd go a day or so without finding one and them I'd find a stack of them again.

It seems to me that if there are infinitely many prime numbers (Euclid) and if for any n there exists arithmetic progressions of length n (Green - Tao,) it should be the case that there are infinitely many arithmetic progressions of n primes for any n. (No, I haven't taken the time to try to prove that, nor am I aware of any proof of it, but it seems to make sense.) It follows from all this that there are, then, infinitely many APs to be double checked (which is all I have managed to accomplich since my original post!)

It only took 3 weeks to find an AP23...

There seemed to be a lot of interest during TDP of the frequency of finds, so I thought I'd post my results for APs (all figures rounded up to the next highest whole number; I've included approximate runtimes, but these will obviously vary by processor. The times listed are for a 1080 Ti at stock clocks).

AP 24+ no data
AP 23 1 : 1,492 ( 1 / 15.64 days)
AP 22 1 : 136 ( 1 / 1.43 days )
AP 21 1 : 50 ( 1 / 12.58 hours )
AP 20 1 : 11 ( 1 / 2.77 hours )
____________