k=5 at 2.8M
 

1962-1968 complete, no primes.

1968-2800 has been already tested by Peter et al., therefore we are now at n=2.8M !

3062-3064 complete, no primes.

3060 to 3062 complete, no primes.

2.8M - 3.0M complete - no primes

taking 3064-3065

grueny

3064-3065 complete - no primes

taking 3065-3066

3065-3066 complete, no primes

Taking 3066 to 3090. This effort has been dormant a year, so I'm putting this block of files on a machine I check about once per month at work.
Peter's k=9 prime at 3.2M inspires me!
-Curtis

Hi Curtis

I marked 3066-3090 as reserved by you. Please note that PSieve released a new version of the 3-4M sieving file about two weeks ago. Our test files are still based on one version before. I'll update our test files next week, or if you prefer you can try to find the file by yourself.

I did fetch the file myself- thanks for the reminder!

3066 to 3090 complete, no primes. I used CUDA-llr for this, so I'm attaching the results in case anyone wishes to double-check any results.

Continuing. I'll post status updates every 10k. I will likely test this k until the FFT value jumps in CUDA.

If anyone has timing interest: 460M is testing these at 2.06 ms/iteration, 6400 sec/test. Windows task manager says it takes about half a core, but it does not slow down 4 LLR clients running on an i7 (less than 5% change). Even 7 other threads on the i7 leaves the CUDA client within 1% of full speed.